Modelling Driver Behaviour in Automotive

Environments
Carlo Cacciabue (Ed.)

Modelling Driver
Behaviourin Automotive
Environments
Critical Issues in Driver Interactions with Intelligent Transport Systems

~ Springer
P.Carlo Cacciabue,Ph.D.
EC IRC -IPSC, Italy

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Control Number: 2006937872

ISBN-10: 1-84628-617-4 e-ISBN-I0: 1-84628-618-2

ISBN-13: 978-1-84628-617-9 e-ISBN-13: 978-1-84628-618-6

Printed on acid-free paper

Springer-Verlag London Limited 2007

Whilst we have made considerable efforts to contact all holders of copyright material contained in this
book, we may have failed to locate some of them. Should holders wish to contact the Publisher, we will
be happy to come to some arrangement with them.

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as
permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced,
stored or transmitted, in any form or by any means, with the prior permission in writing of the
publishers, or in the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms should be
sent to the publishers.

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of
a specific statement, that such names are exempt from the relevant laws and regulations and therefore
free for general use.

The publisher makes no representation, express or implied, with regard to the accuracy of the infor-
mation contained in this book and cannot accept any legal responsibility or liability for any errors or
omissions that may be made.

9 8 7 6 543 2 1

Springer Science+Business Media

springer. com
Contents

Editorial .................................................................................. viii

List of Contributors. ................................................................... XII

Chapter 1. International Projects and Actions on Driver Modelling

1. Modelling Driver Behaviour in ED and International Projects. .......... 3

Maria Panou, Evangelos Bekiaris and Vassilis Papakostopoulos

2. TRB Workshop on Driver Models: A Step Towards

a Comprehensive Model of Driving? .......................................... 26
Delphine Cody and Timothy Gordon

3. Towards Monitoring and Modelling for Situation-Adaptive

Driver Assist Systems. ........................................................... 43
Toshiyuki Inagaki

Chapter 2. Conceptual Framework and Modelling Architectures 59

4. A General Conceptual Framework for Modelling Behavioural

Effects of Driver Support Functions. .......................................... 61
Johan Engstrom and Erik Hollnagel

5. Modelling the Driver in Control................................................ 85

Bjorn Peters and Lena Nilsson

6. From Driver Models to Modelling the Driver: What Do

We Really Need to Know About the Driver? 105
Oliver Carsten

v
VI Contents

Chapter 3. Learning and Behavioural Adaptation 121

7. Subject Testing for Evaluation of Driver Information Systems

and Driver Assistance Systems - Learning Effects and
Methodological Solutions................................................... .... 123
Klaus Bengler

8. Modelling Driver's Risk Taking Behaviour.................................. 135

Wiel Janssen

9. Dealing with Behavioural Adaptations to Advanced Driver

Support Systems. .................................................................. 147
Farida Saad

Chapter 4. Modelling Motivation and Psychological Mechanisms 163

10. Motivational Determinants of Control in the Driving Task................ 165

Ray Fuller

11. Towards Understanding Motivational and Emotional Factors

in Driver Behaviour: Comfort Through Satisficing......................... 189
Heikki Summala

12. Modelling Driver Behaviour on Basis of Emotions and Feelings:

Intelligent Transport Systems and Behavioural Adaptations.............. 208
Truls Vaa

Chapter 5. Modelling Risk and Errors 233

13. Time-Related Measures for Modelling Risk in Driver Behaviour....... 235

Richard van der Horst

14. Situation Awareness and Driving: A Cognitive Model..................... 253

Martin Baumann and Josef F. Krems

15. Driver Error and Crashes ........................................................ 266

Dianne Parker

Chapter 6. Control Theory Models of Driver Behaviour 275

16. Control Theory Models of the Driver......................................... 277

Thomas Jiirgensohn

17. Review of Control Theory Models for Directional

and Speed Control................................................................. 293
David H. Weir and Kevin C. Chao
 Contents VB

Chapter 7. Simulation of Driver Behaviour 313

18. Cognitive Modelling and Computational Simulation

of Drivers Mental Activities. .................................................... 315
Thierry Bellet, Beatrice Bailly, Pierre Mayenobe and Olivier Georgeon

19. Simple Simulation of Driver Performance for Prediction

and Design Analysis.............................................................. 344
P. Carlo Cacciabue, Cristina Re and Luigi Macchi

Chapter 8. Simulation of Traffic and Real Situations 377

20. Real-Time Traffic and Environment Risk Estimation

for the Optimisation of Human-Machine Interaction...................... 379
Angelos Amditis, Aris Polychronopoulos and Evangelos Bekiaris

21. Present and Future of Simulation Traffic Models.. ......................... 400

Fabio Tango, Roberto Montanari and Stefano Mariani

Index...................................................................................... 429
Editorial

The implementation of information technology and automation has been the

driving force of the development of technology in the last decades. At the same
time the presence of humans in control of systems has been kept as the locus of
design principles. Only more recently, fully unmanned technology is beginning to
find its application, in a limited number of domains, such as urban guided trans-
port systems and military aviation. However, even in the cases of totally automatic
systems, it is not possible to avoid the assessment of the human-in-control prin-
ciple, as the overall remote operator of the fully automatic systems remains to
be accounted for in the design and development processes. For these reasons, the
consideration of the user and controller of technologically advanced systems is
one of the most relevant issues for design development, and during production and
implementation processes.
The use of "intelligent" systems implies that the level of autonomy and the
possibility to delegate control processes to technology and automation has im-
proved enormously. The human being has been freed from performing a number
of activities and has progressively been removed from the direct control loop of
systems, in favour of high level decision making processes.
Therefore, in parallel to the development of technology, the need to account for
the behaviour of the human being has progressively evolved from the consideration
of the human-manual-controller to the human-supervisor of processes, procedures
and performances of automatic control systems. The evaluation of behavioural
performance has been replaced by the analysis of cognitive and mental processes.
In other words, the demand for modelling manual and behavioural activities has
been replaced and combined with the need for modelling cognition. This is the
requirement that has mostly affected the development of new technologies and
interfaces in modem control devices.
This feature of design and development of technological systems is common to
many different domains, from energy production, chemical and process industry,
transport and health care. In this scenario, the automotive transport is the domain
mostly affected by the need to consider the multiplicity of human behaviour,
as it presents the highest possible variety of operating environments, of human

viii
 Editorial ix

behaviours, and it offers many different technological solutions for all different
control processes.
In reality, many automation controls currently applied to vehicles already
contain models of a certain complexity of cognition and behaviour, based mostly
on dynamic manifestation of control operations. The "automatic gearbox" of cer-
tain vehicles is a typical example of this type of control systems, which adapt
dynamically and independently to different "driving styles", measured through
intrinsic evaluation of behavioural variables, such as rate of accelerator pressure,
overall speed, etc. Another example is the system that manages the availability
of in vehicle information systems (IVIS), such as telephones or radios. In this
case, certain IVIS managers adapt to the environmental situations, by inhibiting
or discouraging the use of certain IVIS in risky situations.
The models of cognition and behaviour that are implemented in such types of
vehicle control systems are naturally elementary from the cognitive point of view.
However, this shows that this industrial field, both in terms of vehicle integrators
and original equipment manufacturers, needs to apply at design level and integrate
at implementation level adequate models of driver behaviour. These models are
equally important for academic and research purposes, where more complex and
varied solutions can be proposed and studied in relation to theoretical paradigms of
different nature and targets. Another area where modelling of driver behaviour is
essential is the transport safety authorities and regulators, where the consideration
of driver performance becomes essential in setting standards and rules governing
new and future regulations of vehicle control systems, road infrastructures and
traffic management. Similarly, models of drivers are necessary for the study of
accidents and investigation of root causes.
The availability of models and paradigms of driver behaviour at different levels
of complexity and development is therefore quite obvious according to the field of
application.
This book offers to the reader the possibility to assess different approaches and
considerations in relation to driver behaviour modelling, resulting from different
fields. Indeed, the authors of the different manuscripts come from the industrial
area, both car and original equipment manufactures and integrators, from the re-
search and academic fields and from national and international regulators and
automotive transport authorities.
More in detail, Chapter 1presents the ongoing activities in International Projects
and Actions on Driver Modelling. In particular, the European sustained research
Projects carried out over the last decades and presently under development are
reviewed in the paper by Panou, Bekiaris, and Papakostopoulos. Similarly, the US
research actions on driver models and a recently held workshop on these issues
are discussed in the paper by Cody and Gordon. The last paper of this Chapter,
by Inagaky, also revises the actions in Japan on driver modelling, focusing on
monitoring and modelling situation-adaptive driver assistance systems.
After this initial review, more specific subjects are dealt with, beginning with
the existing Conceptual Frameworks and Modelling Architectures (Chapter 2)
that sustain the development of specific models of driver behaviour. In all three
x Editorial

papers, a short historical review of paradigms and architectures for considering

the human element in a Driver-Vehicle-Environment perspective is performed. The
three papers then focus on a specific modelling architecture that enables the reader
to consider very high level structures of interaction, primarily in the papers by
Engstrom and Hollnagel, and by Peters and Nilsson, and more practical and field
focused architectures in the paper by Carsten.
Chapter 3, contains a overall discussion and overview of one of the major issues
that affect driver behaviour modelling: Learning and Behavioural Adaptation. This
issue is dealt with from different perspectives, beginning with the view on Testing
for Evaluation of driver information systems and driver assistance systems, by
Bengler. The crucial issue of risk taking and risk perception is discussed in the
work of Janssen. The Chapter is then completed by the paper of Saad, that offers a
wide and distributed review of theoretical stands on adaptation and processes that
affect driver performance.
In Chapter 4 another essential and widely debated aspect of driver behaviour
is tackled: Modelling Motivation and Psychological Mechanisms. The three con-
tributors to this Chapter offer different and, in some cases, controversial views on
specific formulations and algorithms that can be applied to describe and account for
these critical factors governing driver behaviour. In particular, Fuller concentrates
on determinants of control in the driving task. Summala describes motivational
and emotional factors through the concept of "satisfying". Vaa offers an overview
of his long standing arguments on emotions and feelings.
Chapter 5 deals with Modelling Risk and Errors. The issue of risk management
and risk perception, already dealt with in the Chapter about adaptation, is further
developed here in the paper by Van der Horst, where the specific problem of time-
related measures for modelling risk are discussed in detail. The other two papers
focus more closely on aspects associated with human error. The paper of Baumann
and Krems is dedicated to critical causes of human error. A model based on the
evaluation of Situation Awareness for different driving situations is developed. The
paper by Parker offers instead a wider overview and discussion on the human error
making more correlated to classical and well established theories. The specific
application to the automotive domain demonstrates the possibility to apply such
theories to this domain as it has been done in other cases, e.g., aviation, nuclear
energy etc.
Chapter 6, introduces the second part of the book, dedicated to the review of a
number of numerical simulation and computerised implementation of the theories
and paradigms described in the previous five chapters of the book. In particular,
Chapter 6 discusses the Control Theory Models ofDriver Behaviour. The paper by
Jiirgensohn offers a very comprehensive review of many Control Theory Models
of the Driver that have been developed in the past and are still nowadays very
valuable numerical approaches for describing human machine interactions. The
paper by Weir and Chao also offers a review of control theory approaches, but
then quite readily focuses on a specific control theory approach developed by the
authors over a number of years for describing Directional and Speed Control over
a certain period of time.
 Editorial xi

Chapter 7, concentrates on the problem of the overall Simulation of Driver

Behaviour and representation of the interaction with the vehicle and environment.
The paper by Bellet, Bailly, Mayenobe, and Georgeon describes a computational
simulation of drivers mental activities called COSMODRIVE, based on several
Artificial Intelligence approaches such as the Blackboard architecture and spe-
cific simulation languages that enable the fast and simple description of driver
performance. The paper by Cacciabue, Re and Macchi offers a similar simula-
tion approach, called SSDRIVE, which describes the driver behaviour by means
of Object Oriented Languages and applies the theoretical paradigm described in
Chapter 2 of the book and exploits the power of control theory to describe driver
control actions.
Finally, Chapter 8 deals with the issue of Simulation of Traffic and Real Situa-
tions. The need to simulate the driving context is directly associated to the ability of
modem simulations to predict behaviour and driver interactions with the vehicle in
many driving situations. Consequently in order to make the simulation as realistic
as possible, it is necessary that the models that are coupled to driver simulation are
equally detailed and representative of real traffic situations. Two papers complete
this chapter. The paper by Amditis, Polychronopoulos, and Bekiaris concentrates
on real-time traffic and environment risk estimation, while the paper by Tango,
Montanari, and Marzani describes a set of existing simulations of traffic and dis-
cusses the new orientation and perspectives of future models and simulations of
traffic.
The above description of the Chapters demonstrates that the Book can be read as
a whole in order to get a general perspective of the ongoing topics of development
for present generation of automation and intelligent support systems. At the same
time, it offers the overview on open issues characterising research into new areas of
concern for future control perspectives. However, it is equally possible to consider
selected readings of papers and chapters, when the needs of the reader focus on well
defined and specific subjects, such as for example driver adaptation or simulation
of DVE interactions.
The quality and completeness of this Book rests primarily on the excellence of
the papers that are included. For this reason, the editor of this book is deeply grateful
to all authors who have diligently, professionally and proficiently collaborated to
its development. The editor has simply acted as catalyser and integrator of ideas
and competences, ensuring that the pieces of this puzzle could come together in a
consistent and coherent vision of the problems under scrutiny. The success of this
endeavour rests eventually on the quality of its content and on the response of the
scientific community to which it is addressed.

P. Carlo Cacciabue
List of Contributors

Angelos Amditis Oliver Carsten

I-SENSE Group, University of Leeds,
Institute of Communications and Institute for Transport Studies,
Computer Systems, Leeds LS2 9JT, UK.
Athens, Greece.
Kevin C. Chao
Beatrice Bailly Dynamic Research, Inc.,
INRETS- LESCOT, Torrance, CA, USA.
Bron cedex, France.

Martin Baumann Delphine Cody

Chemnitz University of Technology, UC Berkeley,
Department of Psychology, California PATH,
Chemnitz, Germany. Richmond, CA, USA.

Evangelos Bekiaris Johan Engstrom

Hellenic Institute of Transport, Volvo Technology Corporation,
Athens, Greece. Goteborg, Sweden.

Thierry Bellet Ray Fuller

INRETS-LESCOT, Trinity College Dublin,
Bron cedex, France. Department of Psychology,
Dublin 2, Ireland.
Klaus Bengler
BMW Group,
Forschung und Technik, Olivier Georgeon
Munich, Germany. INRETS-LESCOT,
Bron cedex, France.
P. Carlo Cacciabue
EC, Joint Research Centre, Timothy Gordon
Institute for the Protection and University of Michigan,
Security of the Citizen, Transportation Research Institute,
Ispra (VA), Italy. Ann Arbor, MI, USA.

xii
 List of Contributors Xl11

Erik Hollnagel Roberto Montanari

Ecole des Mines de Paris, University of Modena and Reggio
Pole Cindyniques, Emilia,
Sophia Antipolis Cedex, France. Dipartimento di Scienze e Metodi
dell'lngegneria,
Reggio Emilia, Italy.
Toshiyuki Inagaki
University of Tsukuba,
Department of Risk Engineering, Lena Nilsson
Tsukuba, Japan. VTI,
Linkoping, Sweden.
WielJanssen
TNO Defence, Security & Safety,
MariaPanou
BU Human Factors,
Hellenic Institute of Transport,
Soesterberg, The Netherlands.
Athens, Greece.

Thomas Jiirgensohn
HFC Human-Factors-Consult Vassilis Papakostopoulos
GmbH, Hellenic Institute of Transport,
Berlin, Germany. Athens, Greece.

DianneParker
JosefF.Krems University of Manchester,
Chemnitz University of Technology, School of Psychological
Department of Psychology, Sciences,
Chemnitz, Germany. Manchester M13 9PL, UK.

LuigiMacchi
BjornPeters
EC, Joint Research Centre,
VTI,
Institute for the Protection and
Linkoping, Sweden.
Security of the Citizen,
Ispra (VA), Italy.
Aris Polychronopoulos
Stefano Marzani I-SENSE Group,
University of Modena and Reggio Institute of Communications and
Emilia, Computer Systems,
Dipartimento di Scienze e Metodi Athens, Greece.
dell'lngegneria,
Reggio Emilia, Italy.
Cristina Re
EC, Joint Research Centre,
PierreMayenobe Institute for the Protection and
INRETS-LESCOT, Security of the Citizen,
Bron cedex, France. Ispra (VA), Italy.
xiv List of Contributors

Farida Saad Truls Vaa

INRETS-GARIG, Institute of Transport Economics
Champs-sur-Marne, France. (T0I),
OSLO, Norway.
HeikkiSummala
Richard van der Horst
University of Helsinki, Department of
TNO Defence, Security & Safety,
Psychology,
BU Human Factors,
Helsinki, Finland.
Soesterberg, The Netherlands.

FabioTango DavidH. Weir

Centro Ricerche Fiat (C.R.F.) Dynamic Research, Inc.,
Orbassano (TO), Italy. Torrance, CA, USA.
 I
International Projects and Actions on
Driver Modelling
1
Modelling Driver Behaviour in
European Union and International
Projects
M. PANOU, E. BEKIARIS AND V. PAPAKOSTOPOULOS

1.1 Introduction
Human (or operator) modelling has been an extensive area of research in many
application areas, such as artificial intelligence, aviation, probabilistic risk as-
sessments, system safety analysis and human performances in working contexts
(Cacciabue et al., 1993; Baron et al., 1980). Still, human behaviour is fairly con-
textual and substantially different from one person to another. Thus, the initial
linear models have been gradually replaced by nonlinear and even probabilistic
models, based upon artificial intelligence (AI) principles, such as artificial neural
networks or genetic algorithms. This becomes even more intrigued if we consider
a complex behavioural task such as vehicle driving.
The traffic system as a whole can be seen as being composed of three interactive
parts: vehicles, road users and the road environment. Any traffic situation is the
result of the interaction between these three systems. Normally, the traffic situation
develops as planned, but, in certain circumstances the resulting interaction will
result in a critical situation or in a crash.
The driver is a critical component of the traffic system. Attempts have been
made to estimate the importance of the driver as an accident cause (Evans, 1985).
It has been estimated that road user factors are the sole or contributory factors in
a great majority of road crashes.
There is no generally accepted model of the complete driving task. There are
detailed descriptions focusing on perception and handling aspects and reporting
what drivers really do in every possible ('normal') situation from the beginning
to the end of a journey (see McKnight and Adams, 1970). There are also more
analytical approaches focusing on driver behaviour in relation to task demands,
with the purpose of trying to explain and understand the psychological mechanisms
underlying human behaviour (Rasmussen, 1984; Michon, 1985).
Usually, car driving is described as a task containing three different levels of
demands. At the strategic level, the general planning of a journey is handled. For ex-
ample, the driver chooses the route and transportation mode and evaluates resulting
costs and time consumption. At the tactical level, the driver has to exercise ma-
noeuvres, allowing him/her to negotiate the 'right now' prevailing circumstances,

3
4 Panou, Bekiaris and Papakostopoulos

for instance, turning at an intersection or accepting a gap. Finally, at the control

(stabilisation) level the driver has to execute simple (automatic) action patterns,
which together form a manoeuvre, for example, changing the gear and turning the
wheel.
The demands imposed on the driver are met through his/her driving behaviour.
Also, the performance of the driving task is usually assigned to three different
levels: knowledge-based, rule-based and skill-based behaviour. Skill-based be-
haviour is described as data-driven, meaning that skills are performed without
conscious control and use of attention resources. They are immediate and effi-
cient. Rule-based behaviour on the other hand occurs under conscious control and
requires attention. Therefore, it is less immediate and efficient. Knowledge-based
behaviour involves problem solving and is relevant when it is not given how to act
in a specific situation. Thus, an important aspect of knowledge-based behaviour is
that reasoning is required.

1.2 Evaluation of Driver Behaviour Models

Analysing the driving task requires consideration of the dynamic interaction be-
tween drivers and the traffic system. Driver-specific factors include performance
aspects, individual dispositions and transient driver states. Driver behaviour mod-
els attempt to formalise the complex relation between the driver and the traffic
system.

1.2.1 Michon's Hierarchical Control Model

Michon (1985) proposed a simple two-way classification of driver behaviour mod-
els: One dimension distinguished between behaviour, i.e., input-output-oriented
models and internal-state-oriented models. The second dimension differentiates
between functional models and taxonomic models, where model components do
or do not interact, respectively.
According to Michon (1985), all models lack in one or more respect: they are
generally bottom-up controlled, internal models. Corresponding top-down pro-
cesses are hardly specified or they tend to be too simplistic. Michon regards cogni-
tive process models as the most encouraging step towards a valid model of driver
behaviour because these types of models combine elements of driving task analy-
sis with an information-processing approach. Therefore his Hierarchical Control
Model subdivides the driving task into three coupled and hierarchically ordered
levels, namely the strategic, the manoeuvring and the control levels. Adapting this
model, to incorporate the GADGET fourth level (see next section), one more level
is added, i.e., the 'behaviour level' (Fig. 1.1).
The strategical level includes trip planning, route choice and other general
principles including time constraints. This level is little involved in actual driv-
ing. However, it sets criteria for factors at the lower levels, like speed control
and associated subjective risk levels. At the manoeuvring level, drivers interact
 1. Modelling DriverBehaviour in ED and International Projects 5

Way of living Infinite

General plans Long

Environmental Controlled
input sees
action patterns

Environmental Automatic action 1

input 1000 sees
patterns

FIGURE 1.1. The hierarchical structure of the driving task (adapted from Michon, 1985).

with the traffic system. The control level, finally, refers to basal car control
processes. Although a dynamic relationship between the concurrent activities
is assumed, the different levels require different types of information: The
strategical level is mainly top-down (knowledge) controlled. The manoeuvring
and control levels require in addition bottom-up (data) input from the traffic
environment.
Closely associated to Michon's hierarchical model of driver behaviour is Ras-
mussen's division of operative behaviour into three levels: skill-based behaviour
refers to automatic procedures, rule-based behaviour to application of learned
rules and knowledge-based behaviour to conscious problem solving (Rasmussen,
1984). Skill-based behaviour is applied in Michon's model mainly at the control
level in the form of automatic action patterns. Ranney (1994) relates Michon's
control hierarchy to Rasmussen's taxonomy of operative behaviour. Skill-based
behaviour is applied in all familiar situations. Rule-based behaviour dominates
during standard interactions with other road users as well as in some rare situa-
tions like driving a new car, where automatic routines have to be transferred to a
new system. Knowledge is applied when driving in unfamiliar traffic networks,
in difficult environmental conditions or when skills are not fully developed as in
novice drivers.

1.2.2 The GADGET-Matrix: Integrating Hierarchical

Control Models and Motivational Models ofDriver
Behaviour
Motivational models of driver behaviour 'propose a general compensatory mecha-
nism whereby drivers adjust their driving (e.g. speed) to establish a balance between
what happens on the road and their level of acceptable subjective risk (Ranney,
1994). An important assumption of motivational models is that drivers establish a
6 Panou, Bekiaris and Papakostopoulos

constant level of risk by activating risk-compensation mechanisms when a subjec-

tive threshold is exceeded (e.g., Summala, 1988). The opportunity to compensate
risks by adjusting subjective risk levels indicates that drivers' personal motives
are as well a crucial factor for safe driver behaviour. For this reason, a fourth level
corresponding to individual dispositions has been added to hierarchical control
models of driver behaviour within the European project GADGET (Christ et aI.,
2000). The new level refers to personal preconditions and ambitions in life, and
as such has the highest priority inside the matrix because such dispositions heav-
ily influence driving decisions at lower levels. The four levels of the so-called
GADGET-Matrix are as follows (Table 1.1):

(a) Goals for life and skills for living: An individual driver's attitudes, lifestyle,
social background, gender, age and other personal preconditions that might
influence driving behaviour and accident involvement.
(b) Driving goals and context: Strategical planning of a trip; the focus is on why,
where, when and with whom one is driving.
(c) Mastery of traffic situations: Actual driving in a given context, resembles
Michon's manoeuvring level.
(d) Vehicle manoeuvring: Overlaps despite a different terminology with Michon's
car control level. The focus is on the vehicle, its construction and how it is
operated.

A safe driver has, however, not only developed skills but also knowledge about
his/her own abilities, preconditions and limits. Experienced drivers have, in addi-
tion, cognitive driving skills, such as anticipation and risk perception. In order to
cover these higher-order aspects of driver behaviour, vertical columns are added to
the so far horizontal structure of hierarchical control models (see Table 1.1). The
columns of the GADGET-Matrix are as follows:

(a) Knowledge and skills: Routines and information required for driving under
normal circumstances.
(b) Risk-increasing factors: Aspects of traffic and life associated with higher risk.
(c) Self-assessment: How good the driver reflects his/her own driving skills and
motivations.

Levels and cells of the GADGET-Matrix are not mutually exclusive - there is
large vertical as well as horizontal overlap due to the complex and cyclic nature
of the driving task, where subtasks usually have to be carried out in parallel at
different levels (e.g., routing, turning left, gap acceptance, speed control, steering,
braking, etc.).

1.2.3 DRIVABILITY Model

The most recent evaluation in driver modelling concerns the notion that driver
behaviour is not necessarily static, but evolves dynamically with time, as well
as is context-related. It is subjected not only to permanent but also temporary
 1. Modelling Driver Behaviourin ED and International Projects 7

TABLE 1.1. The GADGET-matrix (adaptedfrom Hatakka et aI., 1999).

Knowledge and skills Risk-increasing factors Self-assessment
Awareness about Risky tendencies like Awareness of
relation between personal acceptance of risks impulse control
tendencies and driving high level of sensation risky tendencies
skills seeking dangerous motives
lifestyle/life situation complying to social risky habits
peer group norms pressure
motives use of alcohol and drugs
personal values

Awareness about Risks associated with Awareness of

effects ofjourney goals physical condition personal planning skills
planning and choosing (fitness, arousal, alcohol, typical driving goals
routes etc.) alternative transport
effects of social pressure purpose of driving modes
by passengers inside the driving environment
car (rural/urban/highway)
social context and
company

Knowledge about Risks associated with Awareness of

traffic regulations wrong expectations strong and weak points of
traffic signs vulnerable road users manoeuvring skills
anticipation violations subjective risk level
communication information overload subjective safety margins
safety margins unusual conditions
inexperience

Skills concerning Risks associated with Awareness of

control of direction and insufficient skills strong and weak points of
position environmental conditions car control skills
vehicle properties (weather, friction, etc.)
physical phenomena car condition (tyres,
engine, etc.)

contributors, which mayor may not be independent. The DRIVABILITY model

(Bekiaris, Amditis, Panou, 2003) introduced as most important the following ones:

1. Individual resources, namely physical, social, psychological and mental con-

ditions of the specific driver. Physical conditions include motor, sensoric and
coordination functions. Mental status depends also on the actual level of stress,
concentration to the task and vigilance level.
2. Knowledge/skills level: This refers not only to actual driver training and ex-
perience, but also to generic knowledge, as basic education greatly influences
motivations and behaviour of the driver. This level also considers the self-
awareness of the own skills and it includes all the four levels of the GADGET
model.
8 Panou, Bekiaris and Papakostopoulos

3. Environmental factors: This includes the vehicle status, the existence of traffic
hazards, the weather, road and traffic conditions. The combinations of these
may generate a risky situation , which certainly influences DRIVABILITY.
4. Two common denominators between driver resources and environmental status,
namely workload and riskawareness.
The two intermediate factors between driver resources and the environment,
namely workload and risk assessment, are among the key issues in order to under-
stand and analyse driving performance. Risk awareness depends on three major
contributors:
1. Risk perception, namely the ability to understand/recognise the specific risk at
the specific time moment.
2. Level of attention, the ability to spot the risk in time.
3. Possible external support so as to spot the risk in time, i.e., by advanced driver
assistance systems (ADAS).

In contrast to the risk awareness level, which is rather discrete and may change
arbitrarily, the other factor, workload, is continuous and evolves with time. Even
temporary input, i.e., use of mobile phone, may have high impact on workload
for limited time periods . The major contributors to DRIVABILITY are depicted
in Fig. 1.2.

FIGURE 1.2. Contributors to DRIVABILITY.

 1. Modelling Driver Behaviour in ED and International Projects 9

The contributors shown in Fig. 1.2 are combined in a mathematical formula,

which comprises the DRIVABILITY index (DI) of each individual driver at any
given moment in time. Considering that the individual resources are the most
significant contributor, the knowledge/skills and workload are of equal importance,
while environment and risk awareness are third in importance, the overall DI is
calculated through the following empirical formula:
KSI WI EFI + RAI
DI=IRlx - x - x ----
226
where IRI is the individual resources index, KSI the knowledge/skills index, WI the
workload index, EFI the environmental factors index and RAI the risk awareness
index.
The DRIVABILITY model contributors validity are being proved within the
AIDE IP, where five modules are being developed, as components of one model,
that monitor whether the driver is engaged and/or distracted by a secondary task
and his/her availability/unavailability (related to the contributor V of Fig. 1.2),
his/her inattention/fatigue (related to contributor IV of Fig. 1.2), his/her personal
characteristics (related to the contributors I and II of Fig. 1.2) as well as the moni-
toring of the traffic and environment (related to the contributor III of Fig. 1.2). In a
few words, the Cockpit Activity Assessment (CAA) module in AIDE is intended
to monitor the activities of the driver to detect workload by visual distraction,
cognitive distraction, and signs of lateral manoeuvring intent. Furthermore, the
Driver Availability Estimator (DAE) module aims to assess the driver's 'level of
availability/unavailability' to receive and process information, according to the
requirements of the primary driving task (depending on the nature of the road in-
frastructure, the goal followed at this time, the current driving actions carried out,
etc.). The Driver State Degradation (DSD) module intends to detect and to diagnose
in real-time the driver hypo-vigilance state due to drowsiness and sleepiness situa-
tions, giving an indication about the driver's ability to execute the driving task. The
Driver Characteristics (DC) module personalises the warning and/or information
provision media, timing and intensity according to driver's profile (experience,
reaction time, average headway, etc.), explicit and implicit preferences. The Traf-
fic and Environment Risk Assessment (TERA) monitors and measures activities
outside the vehicle in order to assess the external contributors to the environmental
and traffic context and also to predict the driver's intention for lateral manoeuvre
(Boverie et al., 2005).

1.3 Driver Behaviour Adaptation Models and Their

Relation to ADAS
ADAS are currently being developed and installed within vehicles at an increasing
rate. These systems aim to improve driving safety by automating aspects of the
driving process, through information, warnings and support to the driver, hence
reducing driver workload.
10 Panou, Bekiaris and Papakostopoulos

TABLE 1.2. Driver behaviour issues when introducing ACC (Bekiaris et al., 2001).
Short term Long term
Mistrust: distrusting the ACC system Spare capacity: using spare capacity for other
in-vehicle tasks
Over-reliance: relying too much on the ACC system
Brake pedal forces: increasing brake pedal Fatigue: ACC could take over too many
forces driving tasks causing fatigue
Imitation: unequipped vehicles imitate Quick approach to vehicle in front: the
equipped vehicles development of new behaviour
Reliance on vehicle in front: vehicle in front Time-headway: driving with smaller
might have poor driving behaviour time-headways
Indication for overtaking: use ACC as an
indication of when to overtake
Overtaking: difficulties with overtaking and being overtaken

Automation can reduce driver workload in areas of decision choice, informa-

tion acquisition and information analysis. This reduction in workload should then
reduce driver error and stress, thereby increasing road safety. However potential
problems exist with the introduction of automated systems. The reduction of men-
tal workload may not always occur under conditions of system failure or when a
user is unfamiliar with the system. Under these conditions, workload may in fact
increase rather than reduce (Stokes et al., 1990). Also, changes in driver skills,
learning and behaviour, which may occur due to the shift in locus of control, may
prove detrimental and therefore predictions as to how drivers will react when lo-
cus of control shifts between the driver and the vehicle are required. Short- and
long-term driver behavioural changes with the use of an advanced cruise control
(ACC) system, from six research projects are summarised in Table 1.2 (Bekiaris
et al., 2001).
The introduction of ADAS, as with any changes to the driving environment,
may lead to changes in driver behaviour. However, the nature of these behaviour
changes in response to changes in the driving environment and has on occasions
proved to be the opposite of that which was intended. Grayson (1996) pointed out
that 'people can respond to innovation and change in ways that are unexpected,
unpredictable, or even wilfully perverse' . For example, Adams (1985) claimed that
the introduction of seat belts in vehicles leads to a perception of greater safety, in
tum leading to drivers increasing their speed on the road.
It has been suggested that improved safety cannot be predicted directly from
the efficiency resulting from improved technology, as people adapt to some kinds
of improved efficiency by taking more risks (Howarth, 1993). The introduction
of safety measures may lead to compensatory behaviours that may reduce the
benefits of the measures being implemented. This phenomenon has most recently
been described as 'behavioural adaptation' (OECD, 1990). However, previous
models explaining the behaviour have termed it as 'risk compensation' and 'risk
homeostasis' .
The most important relevant theories and issues related to driver behavioural
adaptation because of an external stimuli (ADAS introduction in our case) are
summarised below (Bekiaris etaI., 2001) ..
 1. Modelling Driver Behaviour in ED and International Projects 11

1.3.1 Automaticity
Automation refers to the mechanical or electrical accomplishment of work. Some
of the automatic components provided by ADAS act as a substitution for tasks that
humans would otherwise be capable of performing. In other cases, ADAS provides
automatic components, which carry out additional tasks that humans would not
have been capable of but will also assist in the overall driving task.
Introducing automation into the driving task offers several potential advantages
in aspects of both efficiency and safety. For example, the use of dynamic route
guidance systems will assist drivers in taking the most cost-effective route, in
terms of fuel and time, for current traffic conditions. Furthermore, automation can
reduce driver workload in areas of decision choice, information acquisition and
information analysis. This reduction in workload should then reduce driver error
and stress, thus increasing road safety.
Overtrust on a system also brings problems. Drivers may become complacent
and may not detect when a system fails. Drivers are left with a false sense of
security, thereby failing to monitor the system leading to the added disadvantage
of loosing system awareness. Drivers may also lose the opportunity to learn and
retain driving skills. Furthermore, the role of the driver may be reduced to such an
extent that their manual driving skills may degrade. This concept has been termed
'out-of-the-loop familiarity'.
Bainbridge (1987) discusses what she terms the' ironies of automation', which
occur with the changing role of the human in the human-machine relationship
when a system is automated. She points out that the more advanced a system is,
the more crucial the contributions of the human operator become. Automation aims
to eliminate the human factor; however, ironically, the human operator is required
to carry out those tasks that cannot be automated. The human operator is therefore
required to monitor the system and to take over and stabilise the system manually
in situations of system failure. However, as previously discussed manual control
skills deteriorate when using an automated system, leading an experienced user to
become inexperienced.
Bainbridge points out the loss of cognitive skills that an operator using an auto-
mated system, such as ADAS, is likely to suffer. As the retrieval of knowledge from
the long-term memory is dependent on the frequency of use, operators will loose
the benefit of long-term knowledge concerning processes. This practical knowl-
edge can be used to generate strategies in emergency or unusual situations. It is
difficult to teach practical knowledge without experience; it is thus of great concern
that when automating a system this practical experience and the reinforcement by
frequency of use will be taken away from the operator.

1.3.2 Locus of Control

Locus of control is the location of control over a situation or system. The imple-
mentation of ADAS will, in many cases, move the locus of control away from the
driver and instead will lie with the vehicle. It is therefore necessary to consider
12 Panou, Bekiaris and Papakostopoulos

the impact of this change in locus of control both on driver behaviour and also in
terms of safety. The level of control left to the driver must be carefully considered
(e.g., will drivers be given the opportunity to override vehicle decisions and vice
versa?).
Once level of control is decided upon, it is still important to consider the likely
consequences of implementation. Several concerns exist, including those discussed
in the previous section. First, drivers may feel mistrust in the system and experience
problems in handing over control to the vehicle; this factor is dependent on the
confidence a driver has in the capabilities of the system. Secondly, drivers may
overtrust the system; drivers may become dependent on the system. This may be
problematic both in cases where the system fails; the driver may not detect failure
due to reduced monitoring. Also, in situations when the driver is handed back
control, driver's skills and learning may have been diminished due to out-of-the-
loop familiarity.
Finally, driver behaviour and safety during the handover of control between the
driver and the vehicle needs to be considered. De Vos et al. (1997) investigated
safety and performance when transferring control of the vehicle between the driver
and an Automatic Vehicle Guidance (AVG) system. Drivers were found to be able
to leave the automated lane even when high-speed differences and traffic-density
differences between lanes were present. Unsafe interactions were observed in the
scenario of a low-speed manual lane. As expected, increased trust in the reliability
of the system increased driver comfort. However, as headway decreased drivers
were observed to experience greater discomfort, implying that total trust in the
system did not exist.

1.3.3 Risk Homeostasis

One of the most considered and debated models in the area is Wilde's risk home-
ostasis theory. The model bears similarities to earlier models such as that of Taylor,
and Cownie and Calderwood's. Taylor's risk-speed compensation model (1964)
claims simply that the larger the perceived risk, the lower the chosen speed will be.
Cownie and Calderwood (1966) proposed that drivers drive in a way that will main-
tain a desired level of anxiety, leading to the self-regulation of accidents within a
closed-loop; feedback from the consequences of driver decisions will affect future
decisions.
In a like manner, Wilde's risk homeostasis theory holds that drivers have a target
level of risk per unit time that they attempt to maintain. He proposed that drivers
make adjustments that ensure perceived subjective risk is equal to an internalised
target level of risk. The theory asserts that if a driver is provided with additional
safety measures, such as information concerning traffic ahead or the installation of
a seat belt, the driver will exhibit more risky behaviour to compensate and return
to the target level of risk.
Wilde also posited what he named the 'principle of preservation of the accident
rate'. This principle implies that the number of accidents within a given population
is dependent solely on the number of accidents that population is willing to tolerate.
 1. Modelling Driver Behaviour in ED and International Projects 13

1.3.4 Risk Compensation

As with other risk compensation theories, Naatanen and Summala propose a
zero-risk hypothesis, stating that drivers normally avoid behaviour that elicits
fear or anticipation of fear. Avoidance behaviour is motivated by subjective risk,
which according to the theorists is not high enough, thereby leading to accidents.
The main addition to the theory is its focus on the drivers' desired action. The
theory contends that driver behaviour is motivated not only by perception, ex-
pectancy and subjective risk, but also by the relative attractiveness and bene-
fits of carrying out a behaviour in a given situation. Furthermore, Naatanen and
Summala (1973) postulate that the motivation of desired action is the most im-
portant route leading to a driver's decision to take action. The model proposes
that a decision-making process occurs, weighing up the motivating and inhibit-
ing factors, before a decision is made for action, such as overtaking. According
to the model, driver adaptation occurs when perceptions concerning motivations
and subjective risk are altered, hence altering the balance of the decision-making
process.

1.3.5 Threat Avoidance

Fuller's (1984) threat-avoidance theory is developed both from Wilde's theory
of risk homeostasis and the zero-risk model of driver behaviour proposed by
Naatanen and Summala. The theory presupposes that drivers opt for zero risk
of accident and that they make avoidance, competing or delayed avoidance re-
sponses depending on a wide number of factors. These factors are the rewards and
punishments associated with the response, the accuracy of discriminative stimuli
recognition, the subjective probability of a threat, the effectiveness of avoidance
responses and finally the driver's level of arousal. The theory differs from the
other theories in that it is not based on a motivation variable. Instead it views
the driving task as involving learned avoidance responses to potentially aversive
stimuli.
Fuller argues against the presupposition that drivers are capable of monitoring
the probability of an accident. Instead, he proposes that drivers consider the subjec-
tive probability (or likelihood) of an accident. He proposes that the discriminative
stimulus for a potential aversive stimulus is a projection into the future formulated
through the integration of drivers' perceptions of their speed, the road environ-
ment of the intended path and their ongoing capability. Their expectations of the
threat posed by each of these factors combine, leading to either a discriminative
stimulus or no discriminative stimulus. This model also considers the influence
of rewards and punishments. He proposes that when a discriminative stimulus is
detected, the anticipatory avoidance response is not only determined by the sub-
jective probability of expected threat, but also by the rewards and punishments
associated with the various response alternatives. The theory also differs from
that of Wilde, and Naatanen and Summala in that it highlights the role of learned
responses.
14 Panou, Bekiaris and Papakostopoulos

1.3.6 Utility Maximisation

The utility maximisation model proposed by O'Neill (1977) assumes that the
driver has certain stable goals and makes decisions to maximise the expected
value of these goals. Some of these goals are achievable more effectively through
risk-taking behaviour, for example, speeding to save time or gain social status.
These motivating factors are counteracted by the desire to avoid accidents as
well as by fear of other penalties such as speeding tickets. Balancing goals with
the desire to avoid accidents therefore derives driving behaviour choice. O'Neill
claims that the balance, which affects the decision made, is shifted when a safety
measure is introduced. An assumption made by the theory, which has been ques-
tioned (OECD, 1990), is that the driver is 'rational'. In other words, the driver
is an accurate judge of the accident probability resulting from each mode of
behaviour.
Blomquist (1986) also presented a utility maximisation model, which claimed
to illustrate that risk compensation is a natural part of human behaviour when in-
dividuals pursue multiple goals with limited resources. He claimed that drivers
choose target levels of accident risk, based on the perceived net benefits of
safety effort. Again the model proposes that, under plausible conditions, a change
in safety, which is beyond driver control, causes a compensatory change in
driver effort in the opposite direction. Blomquist (1986) likens this theory to
Wilde's theory of risk homeostasis in that utility maximisation focuses on the
choice of safety goals and risk homeostasis focuses on maintenance of those
goals.

1.3.7 Behavioural Adaptation Formula

Evans (1985) proposes a human behaviour feedback parameter by which the
actual safety change in traffic systems is related to that which was expected.
A mathematical representation of the process dictates that feedback can occur
through physical changes to the system, adjustments in user behaviour for per-
sonal benefits and adjustments in behaviour to re-establish previous risk levels.
Evans suggests that human behaviour feedback is a pervasive phenomenon in
traffic systems, which may greatly influence the outcome of safety measures to
the extent that in some cases the opposite effect to that which was intended may
occur.
In a review of driving behaviour theories, Michon (1985) commented that al-
though behavioural adaptation theories explain the behaviour, and the motivations,
attitudes and factors affecting that behaviour are detailed, the actual processes
by which the behaviour occurs are not explained. Similarly, an OECD report
found the models to be vague. The report states that risk compensation theo-
ries do not explain how or which cognitions lead to the expected compensation
of objective risk. The report also points out the problem of how the objective
risk can be accurately assessed by the driver; again models fail to explain this
process.
 1. Modelling Driver Behaviour in ED and International Projects 15

1.4 Use of Driver Behaviour Models in EU and International

Projects

The field of application of such models in ED projects is vast and is gaining pace
over the last decades. Rather than attempting to meticulously cover this extensive
area, we will provide an in-depth overview of characteristic examples, showing
the related difficulties as well as benefits in applying such models.

1.4.1 Driver Models Use for Driver Training and Assessment

Indeed, many of the driver models have been developed, not aiming at driver
behaviour and support but at facilitating better, theory-based, driver training model.
A characteristic example is the GADGET-Matrix developed with the EC project
GADGET in order to structure post-license driver education.
This model as well as different extension of the Michon model have been used
as basis in nearly all recent EC projects dealing with driver training. In TRAINER
project (GRDI-1999-10024), the different layers of the GADGET-Matrix have
been further detailed and correlated to the problems of novice drivers, following
a relevant accident analysis end experts opinion survey. This work resulted in the
adapted GADGET-Matrix that correlates key subtasks of the different GADGET
layers to the needs of driver trainees, with support by multimedia tools and/or
driver simulators (Table 1.3). Thus, the development of appropriate new training
tools and scenario was based upon the relevant theoretical model of novice drivers'
needs.
Another training (in fact re-training) and assessment application is that of AG-
ILE project (QLRT-2001-00118), regarding the assessment of driving ability and
eventual re-training of elderly drivers. In this project, a mapping has been at-
tempted of the age-related deficits and benefits to the different levels and cells
of the GADGET-Matrix (Breker S. et al., 2003). The second level of the adapted
GADGET-Martix is presented in Table 1.4.
This resulted in the prioritisation of specific driving scenaria, where elderly
drivers would need more thorough assessment and/or support (re-training or
aiding).

1.4.2 Evaluation ofDriver Models' Use for Safety Aids

1.4.2.1 Use of Seat Belts
Evans (1982) observed that unbelted drivers drive at higher speeds and with smaller
headways in comparison to drivers wearing belts. This evidence supports the theory
that an adaptation in behaviour has occurred; however, the reduction in risk-taking
behaviour does not support the risk compensation hypothesis. In contrast, Streffand
Geller (1988) found that go-kart drivers wearing seat belts drove faster than non-
wearers, suggesting that the seat belt leads to a sense of security, enabling drivers
to feel safe in increasing vehicle speed. The experimental validity is questionable
16 Panou, Bekiaris and Papakostopoulos

TABLE 1.3. Findings of the analyses made in TRAINER D2.1.

Experts' proposals
Literature Existing
survey training Multimedia Simulator

Starting 0 X
Shifting gears 0 X
Accelerating/decelerating 0 X
Steering/lane following 0 X
Speed control 0 X
Braking/stopping 0 X
Use of new cars control aids (ABS, ACC, etc) X X X
Insufficient skills and incomplete automation X
Realistic self-evaluation X
Following X 0 X
Overtaking/Passing X 0 X X
Entering and leaving the traffic X 0 X
Tailgating X X
Lane changing X 0 X X
Scanning the road (eye cues) X 0 X X
Reacting to other vehicles X OX X
Reacting to pedestrians X X X
Parking 0
Negotiating intersections X 0 X X
Negotiating hills/slopes X 0 X
Negotiating curves X 0 X X
Road surface (skid, obstacles) X OX X
Approach/exit of motorways X 0 X
Turning off/over 0
Railroad crossings, bridges, tunnels 0
Reacting to traffic signs and traffic lights X 0 X X
Reacting to direction signs (including in-car X
devices)
Emergency brake OX
Urban driving 0 X X
Rural driving 0
Convoy driving X
Motorway driving OX X
Weather conditions X OX X
Night driving X OX X
Insufficient skills and incompletely X X X
automation
Information overload X X X
Insufficient anticipating skills and wrong X X X
expectations
Risky driving style X X X
Realistic self-evaluation X X X
Awareness of personal driving style X X X
Determination of trip goals, route and modal
choice
Preparation and technical check 0 X
Safety issues X 0 X
Maintenance tasks 0 X
 1. Modelling Driver Behaviour in ED and International Projects 17

TABLE 1.3. (Continued)

Experts' proposals
Literature Existing
survey training Multimedia Simulator

Intemationallegislation X
First aid (}X X
Economic driving x o X
Driver's condition (stress, mood, fatigue) X X X
Motives for driving X
Awareness of personal planning skills
Awareness of typical driving goals and risky X
driving motives
Knowing about the general relations X
between lifestyle/age/gender and
driving style
Knowing the influence of personal values X
and social background
Knowing about the influence of passengers X X
High level of sensation seeking X
Consequences of social pressure, use of X X X
alcohol and drugs
Awareness of own personal tendencies X
(risky habits, safety-negative motives)

0: the task is trained in all or nearly all European countries as the analysis of the questionnaires
showed; 0: the task is trained only in few or at least one country; X: the driving authorities and
driving instructors questioned indicate that the task is not trained, but should be trained in the particular
country.

because of the generalisation of go-karting to real life driving, as well as due to the
differences in overall behavioural patterns of drivers using seat belts versus those
that do not.

1.4.2.2 Use of Motorcycle Helmet

The retraction of these laws in some states of the USA during the late 1970s made
it possible to compare repeal and non-repeal conditions in a natural environment.
However, results from accident studies suggested that wearing helmets provided
a safety benefit; those states that had revoked laws requiring helmets to be worn
suffered an increase in fatalities (Grayson, 1996). These findings were opposed by
the analysis of Adams (1983), where higher fatality rates were found in those states
that had retained helmet-wearing laws. Adams argued that these results support the
risk compensation theory: motorcycle riders who wore helmets felt less vulnerable
to injury, thereby exhibiting riskier driving behaviours. The findings of Adams in
favour of the risk compensation theory have again been disputed by the work of
Chenier and Evans (1987). In their re-analysis of the USA accident statistics, they
found that fatalities were increased in the states that had retracted compulsory
crash helmet wearing laws. Their results imply that any increase of caution due to
TABLE 1.4. GADGET-Matrix level 2 (Mastery of Traffic Situations): Situation of older drivers (Breker et aI., 2003).
Knowledge and skills Risk factors Self-assessment

Pro Contra Pro Contra Pro Contra

Knowledge about traffic regulations/traffic Risks associated with wrong expectations/vulnerable road Awareness of strong and weak points of
signs/anticipation/communication/safety users/violations/information overload/unusual conditions/inexperience, manoeuvring skills/subjective risk
margins, etc. etc. level/subjective safety margins, etc.
-priority (esp. right before -slower on motorways - interpreting movement of other drivers -larger safety -over estimation
left, right-hand driving) -uniform driving style on -judging other drivers movement margins (headway of one's own
-signalling country roads -junctions/intersections etc.) driving abilities
- reduced perception of -larger gaps, especially - yield right of way -lower risk level -less sensibility
road signs when turning left -right-angle (side) collisions -avoiding complex to changes in
-dementia -less overtaking -left turns (right-hand driving) settings performance
-cataract -less night driving -right turns (left-hand driving) -avoiding give-way and capacities
-diabetes + associated -slower approach of - multiple vehicle accidents or stop junctions - insufficient
-glaucoma junctions -merging onto motorways -avoiding heavy awareness that
-cardiac & cardiovascular -early speed reduction at -risky merging in traffic flow traffic oneself is
condition junctions -running over red -awareness of subject to
-seizure disorders -smooth slow down at -railway crossings difficulties at general
-back pain junctions -too early observation of situations (it might changed) intersections age-related skill
-less driving on low -situation complexity especially in urban settings - general awareness declines
friction -late detection of other road users of age-related skill -dementia
-early observation of -judgement of gaps declines
situations - underestimation of speed of vehicles at higher speeds
- tolerance towards other - problems with reappearing situations when new
road users information becomes available
-problems with interrupting actions when necessary
-problems with situation complexity
-skill decline by compensation
-lateral safety margins
-dementia
-cataract
-diabetes + associated
-glaucoma
-cardiac & cardiovascular condition
- seizure disorders
-back pain
 1. Modelling Driver Behaviour in ED and International Projects 19

the removal of helmets was not great enough to compensate for the loss of safety
benefits that the helmets provide. Overall, research into motorcycle helmet wearing
has been found to provide a safety benefit and offers little support to any of the
behavioural adaptation theories. However, as Grayson (1996) pointed out these
findings are not surprising and more relevant to the theories are those mechanisms
which protect one part of the anatomy and lead to disregard for safety of other
parts of the body.

1.4.2.3 Studded Tyres

Studded tyres have been developed to improve safety in icy and snowy conditions;
they provide better track-holding properties and shorter breaking distances under
these conditions. Evidence from studies investigating the behavioural effects of
fitting vehicles with studded tyres have often been cited in support of both be-
havioural adaptation (OECD, 1990) and risk compensation (Adams, 1985). The
most greatly cited study is that of Rumar (1976). Results from this study indicated
that in icy (low friction) conditions drivers of vehicles equipped with studded tires
drive at faster speeds when negotiating curves in the road. However, it was deter-
mined that this increase in speed does not lead to a reduction in safety. It should
therefore be noted that the observation of higher speeds in itself is inconclusive.
Lund and O'Neill (1986) argue that the increased feedback provided by studded
tyres allows vehicles to be driven at higher speeds without reducing safety. The
Rumar study therefore provides evidence that the behavioural effect of driving
faster with studded tyres reduces the safety benefits; however, an overall increase
in safety is still achieved through their implementation.

1.4.2.4 Antilock Braking Systems

Antilock braking systems (ABS) are a recent safety feature introduced to the ve-
hicle, and studies concerning their effect on driver behaviour are cited to support
some of the theories of behavioural adaptation. ABS are designed to make breaking
distances shorter and to allow vehicles to be steered during breaking manoeuvres.
Rompe (1987) conducted a series of tests to investigate the benefit of these sys-
tems. Results showed that when simulating high-risk manoeuvres drivers without
ABS made 2.4 times more errors. Real life evidence does not support this size-
able predicted benefit and therefore support theories of behavioural adaptation. A
study conducted by Aschenbrenner (1994) is one of the only studies designed to
specifically investigate the risk compensation theory; their hypothesis being that
ABS will fail to reduce accidents despite its technical benefits. The study looked
at two fleets of taxis in Munich: one fitted with ABS and the control one without.
Aschenbrenner concluded that since it was not possible to prove a universal in-
crease in safety, the results indicated the occurrence of behavioural adaptation in
the form of risk compensation. The overall accident rate was unchanged, but incon-
sistencies such as decreases in blameworthy accidents and increases in parking and
reversing accidents for fitted vehicles were observed. These inconsistencies were
20 Panou, Bekiaris and Papakostopoulos

described as indicating a tendency to riskier driving by drivers of fitted vehicles

when reviewed in the GECD report.
More supporting evidence has been provided by a US study (HLDI, 1994).
Findings from this study indicated that the introduction of ABS has failed to reduce
the frequency or cost of insurance claims. However, this does not necessarily imply
that ABS does not provide a safety benefit, as Grayson (1996) points out, the
circumstances under which ABS could prevent accidents are quite rare. Kullgren
(1994) supplied evidence that ABS is effective in reducing accidents. This analysis
of Swedish accident statistics indicated an overall effectiveness of 15% for ABS
vehicles under snowy or icy conditions. It was also found that fitted cars were more
likely to be struck from behind in rear-end accidents.
Evidence for the occurrence of behavioural adaptation is both inconsistent and
conflicting. The GECD report reviewed empirical evidence concerning behavioural
adaptation. It was concluded that behavioural adaptation does occur although not
consistently, and the magnitude and direction of its effects on safety cannot be
precisely stated. The studies reviewed suggested that behavioural adaptation does
not eliminate safety gains from programmes but tends to reduce the size of the
expected benefits.

1.4.3 Driver Models Use for ADAS Design and

Impact Assessment
Few studies have been carried out investigating behavioural effects of future au-
tomated systems, and many of these have revealed the occurrence of negative
behavioural effects.
A study conducted by Winsum et al. (1989) provides support to the theory that
drivers will exhibit behavioural adaptation in response to ADAS. Winsum et al.
suggested that the use of a navigation system, in place of a map, leads to a reduction
in workload, which in tum leads to drivers increasing vehicle speed, implying that
drivers demonstrate behavioural adaptation in response to the implementation of
automated navigation.
Similarly, Forward (1993) reviewed the evidence concerning the effects of dy-
namic route guidance (DRG) systems, concluding that benefits such as reduced
workload and stress exist as do undesirable effects such as increased speed. How-
ever, Forward noted that the real effect of the system cannot be fully comprehended
until its use becomes more extensive.
Focus mainly tends to concentrate on the negative effects of introducing au-
tomation to the vehicle. However automation benefits have also been observed
during evaluation of future systems.
In the case of AWAKE (IST-2000-28062), the DRIVABILITY model has been
employed to design its warning levels and strategy. AWAKE was a project aimed
to develop an unobtrusive and personalized, real-time driver monitoring device,
able to reliably predict driver hypovigilance and effectively and timely warn
the driver. AWAKE has recognised the importance of actual traffic risk level as
well as driver status, type and key environmental factors, and worked towards a
 1. Modelling Driver Behaviour in ED and International Projects 21

TABLE 1.5. Correlation of AWAKE (driver vigilance monitoring and warning system) use
cases and warning strategy with overall DRIVABILITY Index (Bekiaris et aI., 2003).
Values of Overall
DRIVABILITY DRIVABILITY AWAKE warning
AWAKE use cases Indexes Index levels/strategy

Driver is hypovigilant; IRI = 3.5 4 (the system closely No action but

Rural environment, with WI=2 monitors the driver, monitoring system
sufficient traffic density without any action) parameters are
(normal workload); strengthened to
No major environmental EFI= 3 cautionary case
risk identified;
Standard type of driver; KSI = 2
No sign that the driver RAI=3
missed any risk.
Thus DI = 3.5
Driver is hypovigilant; IRI = 3.5 3 (the system provides Driver warning by audio
Urban environment, with WI=2 warning) and visual means
normal traffic density (warning level 1)
(normal workload);
No major environmental EFI= 3
risk identified;
Standard type of driver; KSI= 2
Driver seems to miss RAI= 2
some risks (i.e. rather
small TTC or headway).
Thus DI = 2.9
Driver is hypovigilant; IRI = 3.5 2 (the system intervenes) Driver warning by audio
Highway environment, WI=l,4 and haptic means
with low traffic density (warning level 2)
(low workload); (intervention is
High speed, cautionary EFI = 2 excluded from
case; AWAKE, due to
Standard type of driver; KSI = 2 liability issues)
Driver seems to miss RAI= 2
some risks (i.e. lane
deviation or swerving).
Thus DI = 1.75

multi-stage driver monitoring and driver warning system that takes such param-
eters into account. Table 1.5 correlates the overall DI and the indexes of the
DRIVABILITY contributors to the different AWAKE driver warning levels and
media. It should be noticed that the sensors included in the AWAKEsystem (such as
driver eyelid and steering grip force monitoring, frontal radar, lane recognition sys-
tem, etc.) allow for sufficient, real-time estimation of all DRIVABILITY indexes
(except KSI, which is however included in the system by the driver at its initiation
as the system adapts itself to the driver profile). This is done by storing driver's,
vehicle and environmental data on the system and automatically processing them.
Further processing off-line is also feasible.
22 Panou, Bekiaris and Papakostopoulos

TABLE 1.6. Driver model and rules for implementation.

PIPE drivermodel Typeof process Rules or governing assumptions
Perceptionof signals Sensorialprocess - Haptic
- Visual
-Aural
Interpretation Cognitiveprocess - SimilarityMatching
- FrequencyGambling
Planning Cognitiveprocess - Inference/reasoning
Execution Behavioural process - Performance of selectedactions/iterations

This is indeed one of the very few cases where such a direct relation between
a driver behaviour model and the development of the HMI of an ADAS has been
attempted, and in fact with great success, as the final AWAKE HMI has been rated
as adequate and useful by over 90% of its users.
Finally, within AIDE (IST-1-507674), a new driver model is being developed, at-
tempting to model concurrently the driver, the vehicle and the environment (Panou
et al., 2005). The basic assumption made for the development of the model of the
driver is that the driver is essentially performing a set of actions that are familiar ac-
cording to his/her experience. As the driving process is very dynamic, these actions
are continuously selected from a vast repository of knowledge (knowledge base) by
a diagnostic process. Consequently, the processes of diagnosis and interpretation
of acquired information become crucial for the dynamic sequencing of driver's
activity. The model of the driver adopted is based on a very simple approach that
assumes that behaviour derives from a cyclical sequence of four cognitive func-
tions: perception, interpretation, planning and execution (PIPE). This model is
not sequential as the execution function, i.e., the manifested form of behaviour,
may result from several iterations (cyclical) of the other functions. Moreover, in
agreement with the initial hypothesis, the planning function, is usually bypassed
by the 'automatic' selection of familiar frames of knowledge that are associated
with procedures or sets of several actions aiming at the fulfilment of the goal of a
frame. This function is important as it becomes effective in unknown situations or
in the case of novice drivers, when 'simpler' frames, based on single actions or on
a limited sequence of very simple/familiar actions, are called into play to deal with
the situation. These four cognitive functions can be associated to either sensorial
or cognitive processes and are activated according to certain rules or conditions
(Table 1.6). This model will be utilised in personalising the multi-ADAS system
HMI, in accordance to a particular driver's needs and preferences, and will also
be used in the traffic environment.

1.5 Conclusions
Driving task modelling has started as simple task-layers representation for taxo-
nomic use in driver training and has gradually evolved to dynamic models, which
consider driver behaviour adaptation as well as the impact of the traffic environment
 1. Modelling Driver Behaviour in ED and International Projects 23

and the driving context. The initial list of driver training and assessment projects
that used driver models as their theoretical basis (GADGET, DAN, TRAINER,
AGILE, CONSENSUS, etc.) have been followed by a new generation of projects
that use driver models to assess the impacts of driver support systems (i.e.,
ADVISORS, TRAVELGUIDE) and, more recently, by those that attempt to use the
model parameters for optimal HMI design (AWAKE, COMUNICAR, AIDE, etc.).
Preliminary results have proven that such a correlation is feasible and beneficiary,
but it is far from obvious. The model output has to be evaluated and even modi-
fied by empirical results. Thus, currently the model is being applied and tested in
AIDE, SENSATION and PREVENT Integrated Projects through short- and long-
term testing of drivers. Furthermore, the model can only be at the starting basis
of the design and development process and only influence the actual ADAS HMI
within predefined design boundaries. Nevertheless, we seem to be at the infancy
of a new design principle for driver support training and assessment systems - the
model-based modular and personalised design.

References
Adams, 1.G.U. (1985). Smeed's Law, seat belts and the emperor's new clothes. In
L. Evans and R.C. Schwing (Eds.). Human Behaviour and Traffic Safety. Plenum,
New York.
Aschenbrenner, K.M. and Biehl, B. (1994). Empirical studies regarding risk compensation
processes in relation to anti-lock braking systems. In R.M. Trimpop and GJ.S. Wilde
(Eds.). Challenges to Accident Prevention: The Issue of Risk Compensation Behaviour.
Styx Publ., Groningen, The Netherlands.
Bainbridge, L. (1987). Ironies of automation. In 1. Rasmussen, K. Duncan and 1. Leplat
(Eds.). New Technology and Human Error. Wiley, New York.
Baron, S., Zacharias, G., Muralidharan, R. and Lancraft , R. (1980). PROCRU: A Model
for Analyzing Flight Crew Procedures in Approach to Landing. NASA CR-152397.
Bekiaris, E., Papakonstantinou, C., Stevens, A., Parkes, A., Boverie, S., Nilsson, L.,
Brookhuis, K., Van Wees, K., Wiethoff, M., Damiani, S., Lilli, F., Ernst, A., Heino, A.,
Widlroither, H. and Heinrich, J. (2001). ADVISORS Deliverable 3/8.1v4: Compendium
of Existing Insurance Schemes and Laws, Risk Analysis ofADA Systems and Expected
Driver Behavioural Changes. User Awareness Enhancement, dissemination report and
market analysis and ADAS marketing strategy. ADVISORS Consortium, Athens, Greece.
Bekiaris, E., Amditis, A. and Panou, M. (2003). DRIVABILITY: A new concept for mod-
elling driving performance. International Journal ofCognition Technology & Work, 5(2),
152-161.
Blomquist, G. (1986). A utility maximization model of driver traffic safety behaviour.
Accident Analysis and Prevention, 18,371-375.
Boverie, S., Bolovinou, A., Polychronopoulos, A., Amditis, A., Bellet, T., Tattegrain-Veste,
H., Manzano, 1., Bekiaris, E., Panou, M., Portouli, E., Kutila, M., Markkula, G. and
Angvall A. (2005). AIDE Deliverable 3.3.1: AIDE DVE monitoring module - Design
and development. AIDE Consortium, Toulouse, France.
Breker, S., Henrikson, P., Falkmer, T., Bekiaris, E., Panou, M., Eeckhout, G., Siren, A.,
Hakamies-Blomqvist, L., Middleton, H. and Leue E. (2003). AGILE deliverable 1.1:
Problems of elderly in relation to the driving task and relevant critical scenarios.
24 Panou, Bekiaris and Papakostopoulos

Cacciabue, P.C., Mauri, C. and Owen, D. (1993). Development of a model and simulation
of aviation maintenance technician task performance, International Journal ofCognition
Technology & Work, 5(4),229-247.
Chenier, T.C. and Evans, L. (1987). Motorcyclist fatalities and the repeal of mandatory
helmet wearing laws. Accident Analysis and Prevention, 19, 133-139.
Christ et al. (2000). GADGET final report: Investigations on influences upon driver be-
haviour - Safety approaches in comparison and combination. GADGET Consortium,
Wien, Austria.
Cownie, A.R. and Calderwood, 1M. (1966). Feedback in accident control. Operational
Research Quarterly, 17, 253-262.
De Vos, A.P., Hoekstra, W. and Hogema, I.H. (1997). Acceptance of automated vehicle
guidance (AVG): System reliability and exit manoeuvres. Mobility for everyone. Pre-
sented at the 4th World Congress on Intelligent Transport Systems, Berlin.
Evans, L. (1985). Human behaviour feedback and traffic safety. Human Factors, 27 (5),
555-576.
Evans, L. and Schwing, R.C. (Eds.) (1982). Human Behaviour and Traffic Safety. Plenum,
New York.
Forward, S.E. (1993). Prospective methods applied to dynamic route guidance. In O.MJ.
Carsten (Ed.). Framework for Prospective Traffic Safety Analysis. HOPES Project De-
liverable 6.
Fuller, R. (1984). A conceptualization of driving behaviour as threat avoidance. Ergonomics
27, 1139-1155.
Grayson, G. (1996). Behavioural adaptation: A review of the literature. TRL Report 254.
Transport Research Laboratory, Crowthorne.
Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A and Hernetkoski, K. (1999). Results
of EU-project GADGET. In S. Siegrist (Ed.). Driver Training, Testing and Licensing-
Towards Theory Based Management ofYoung Drivers' Injury Risk in Road Traffic. BFU-
report 40, Berne.
HLDI. (1994). Collision and property damage liability losses ofpassenger cars with and
without antilock brakes. Highway Loss Data Institute Report A-41. HLDI, Arlington,
VA.
Howarth, I. (1993). Effective design: Ensuring human factors in design procedures. In
A. Parkes and S. Franzen (Eds.). Driving Future Vehicles. Taylor and Francis, London.
Kullgren, A., Lie, A. and Tingvall, C. (1994). The effectiveness of ABS in real life accidents.
Paper presented at the Fourteenth International Technical Conference on the Enhanced
Safety of Vehicles, Munich.
Lund, A.K. and O'Neill, B. (1986). Perceived risks and drinking behaviour. Accident Anal-
ysis and Prevention, 18, 367-370.
McKnight, AJ. and Adams B.D. (1970). Driver Education Task Analysis, Vol. 1: Task
Descriptions. Human Resources Research Organization, Alexandria, VA.
Michon, lA. (1985). A critical view of driver behaviour models: What do we know, what
should we do? In L. Evans and R.C. Schwing (Eds.). Human Behaviour and Traffic Safety.
Plenum, New York.
Naatanen, R. and Summala, H. (1973). A model for the role of motivational factors in
drivers' decision making. Accident Analysis and Prevention, 6, 243-261.
OECD. (1990). Behavioural Adaptations to Changes in the Road Transport System. OECD,
Paris.
O'Neill, B. (1977). A decision-theory model of danger compensation. Accident Analysis
and Prevention, 9, 157-165.
 1. Modelling Driver Behaviour in EU and International Projects 25

Panou, M., Cacciabue, N., Cacciabue, P.C. and Bekiaris, E. (2005). From driver modelling
to human machine interface personalisation. Paper presented at the IFAC World Congress,
Prague.
Ranney, T. (1994). Models of driving behaviour: A review of their evolution. Accident
Analysis and Prevention, 26(6), 733-750.
Rasmussen, J. (1984). Information processing and human-machine interaction. In An ap-
proach to cognitive engineering. North Holland, New York.
Rompe, K., Schindler, A. and Wallrich, M. (1987). Advantages of an anti-wheel lock system
(ABS) for the average driver in difficult driving situations. Paper presented at the Eleventh
International Technical Conference on Experimental Safety Vehicles, Washington, DC.
Rumar, K., Berggrund, U., Jernberg, P. and Ytterbom, U. (1976). Driver reaction to a
technical safety measure - Studded tyres. Human Factors, 18, 443--454.
Stokes, A., Wickens, C. and Kite, K. (1990). Display Technology: Human Factors Concepts.
Society of Automotive Engineers, Inc., USA.
Streff, F.M. and Geller, E.S. (1988). An experimental test of risk compensation: Between-
subject versus within-subject analyses. Accident Analysis and Prevention, 20, 277-287.
Summala, H, Lamble, D. and Laakso, M. (1988). Driving experience and perception of
the lead cars braking when looking at in-car targets. Accident Analysis and Prevention,
30(4), 401--407.
Taylor, D.H. (1964). Drivers' galvanic skin response and the risk of accident. Ergonomics,
7, 439--451.
Van Winsum, W., Van Knippenberg, C. and Brookhuis, K. (1989). Effect of navigation
support on driver's metnal workload. In Current Issues in European Transport, Vol. I:
Guided Transport in 2040 in Europe. PTRC Education and Research Services, London.
2
TRB Workshop on Driver Models: A
Step Towards a Comprehensive Model
of Driving?
DELPHINE CODY AND TIMOTHY GORDON

2.1 Introduction
Various disciplines use the same or similar terminology for driver models - vehicle
and traffic engineering, psychology, human factors, artificial intelligence to men-
tion the most common; however, the definition of the term varies not only between
disciplines but even between different researchers within any given discipline. Re-
cent efforts in applied psychology and human factors have emphasised the need
of developing models that can be implemented and used in computer simulation,
hence representing a possible link between these disciplines, and also a chance
to consider the broader picture of driver model within a transportation/traffic sys-
tem. In order to discuss this link, the authors organised a workshop on driver
modelling during the 84th annual meeting of the Transportation Research Board,
Washington, DC. The workshop was attended by 25 researchers from the vari-
ous fields listed above and lead researchers from the United States, Europe and
Asia.
Three objectives were set for this workshop: (i) create a group of driver model
developers and users, (ii) share common experience and reach common definitions
and finally (iii) set a road map for the next generation of driver models. The topics
that were more specifically dealt with were the design and application of cognitive
and driver behaviour models and their integration within a broader simulation
framework. This chapter is divided into three sections. In the first section, we
present the workshop content in more detail and provide a summary of the speakers'
contribution. In the second section, we present a synthesis of these models with
the introduction of a set of dimensions allowing for visualisation of the different
models on a similar scale. The final section presents what the authors believe to
be the critical steps necessary to coordinate efforts towards a new generation of
driving model- a framework where researchers from different fields contribute to
create a comprehensive model of the driving activity.

26
 2. TRB Workshop on Driver Models 27

2.2 Workshop Presentation and Speakers' Contribution

This section comprises two parts. We will first detail each of the themes around
which the speakers organised their presentation and will present summaries of
each of the presentations in the second part.

2.2.1 Workshop Content

The workshop was based on three themes: (i) driver model purpose and applica-
tion, (ii) driver model architectures and implementations and (iii) driver model
calibration and validation. We will now discuss each of these themes.

2.2.1.1 Driver Model Purpose and Application

From a general standpoint, generating driver models can be seen as equivalent
to developing a comprehensive description of scientific knowledge about drivers.
The first step towards making a hierarchy of scientific methods to approach a phe-
nomenon such as the driving activity is to conduct specific studies to observe and
understand the phenomena (e.g., drivers' glances, perception of speed, decision-
making process, drivers' attention control) and the second step is to reconcile the
results of the different studies into a comprehensive description - in other words
a model dynamically linking the results of these different studies to reproduce the
driving activity. Hence, one of the interests of building driver models is to vali-
date the results of studies looking into specific aspects of driving and identify the
aspects that need to be addressed to further the understanding of driving.
Another interest for developing such models resides in their application. Two
types of applications were discussed during the workshop: (i) application to safety
and driver assistance system design and (ii) prediction and evaluation of new
systems on traffic flow. Three presentations fell under the first type, namely, 'In-
Vehicle Information System (IVIS) Model' by Jon Hankey, 'ACT-R Driver Model'
by Dario Salvucci and 'Workload in Driver Modelling' by Jeroen Hogema and
Richard van der Horst. The two other presentations, 'ACME - Driver Model' by
Daniel Krajzewicz and 'FLOWSIM' by Mark Brackstone, addressed the traffic
assessment category.

2.2.1.2 Driver Model Architecture and Implementations

Model architecture can be seen as the blueprint of the model. In that sense, a
simple description of a model (see Fig. 2.1) consists of three elements: inputs,
information processing or behaviour and outputs. This simple architecture allows to
generate parameters of factors that can be used to derive measures of effectiveness
or evaluation.
28 Codyand Gordon

/ Driver Models Information Processing

' Perception

/
Inputs Outputs

~
' Cognitive Processes
' driver assistance systems -Hurnan motor control
Vehicle

I
' other vehicles
'road geometry and signals -Decisions

Driver behavior
Vehicle control
\.
1
FactlJrs/ parameters

Measures of Effectiveness/Evaluation
Risk acceptance
safety
I--

Cras h risk
Efficiency
Workl oa d
Stress

FIGUR E 2.1. Modularhigh level representation of drivermodel structure.

2.2.1.3 Calibration and Validation

Model calibration consists of setting values for parameters which support the sim-
ulation. For example, when a model includes a gap regulation task, parameters that
need to be calibrated are the gaps that the simulated drivers will maintain as well
as conditions for variation of these gaps. The calibration addresses elements of
the driving tasks as well as cognitive processes (e.g., memory decay) or other ele-
ments of the simulation (e.g., vehicle models) . We distinguished two main methods
for calibrating model s. The first method consists of using data and results avail-
able in the scientific community, and the second consists of running specific data
collection. This method can be subdivided into three more categories, depending
on whether the data collected is driver centric (gaze, attention) or vehicle cen-
tric (steering, braking, speed contro l) or traffic centric (lane keeping , range , range
rate) . The use of the first method brings up the issue of standardisation in method
to gather and measure data and the definition of parameters that are derived .
For model validation, there is one method of comparing the model output with
data sets. The variation in the application of the method depends on the type of
data used. For instance, the model output can be compared with collected data,
with on-going behaviour, or with more general data . The comparison of model
output with on-going behaviour raised an interesting question on the nature of
the model outp ut. Is a model providing a trend or should we expect to be able
to predict real-time behaviour? The question was not answered directly during
the workshop. Although some researchers presen t in the audience thought that a
prediction of driver behaviour in real time would be a huge achievement for driver
models, a lot of doubt about the possibility to reach such a level of accuracy was
also expressed.
Clearly, there is fundamental issue resulting from the inherently stocha stic na-
ture of driving . On the one hand, a ' real-time' model should be capable of making
precise and detailed predictions of driver actions as a function of time. On the other
hand, real drivers display a range of decisions, driving styles, levels of attention,
 2. TRB Workshop on Driver Models 29

etc., and so should any realistic driving model; resolving these apparently conflict-
ing viewpoints requires a somewhat deeper concept of what is required for model
validation. At the very least, validation of real-time models must take account for
a wide range of stochastic influences.
Out of the five models that were presented, four are proceeding with more or
less extensive validation and one is still in the process of calibration.

2.2.2 Summaries of the Speakers' Contributions

Five speakers presented their research relative to each of the themes: Jon Hankey,
Virginia Tech Transportation Institute; Dario Salvucci, Drexel University; Daniel
Krajzewicz, Deutsches Zentrum fur Luft - und Raumfahrt e.V. (DLR); Mark Brack-
stone, School of Engineering and the Environment, University of Southampton,
and Richard van der Horst, TNO Human Factors. All the five summaries are di-
vided into three parts: (i) model purpose description, (ii) architecture description
and (iii) calibration and validation issues. The architectures that were presented are
briefly explained for each of the model. The reader may refer speakers' publica-
tions on their model for more thorough descriptions. The models are also discussed
based on the following factors: (i) simulation scale, along two dimensions, i.e., the
unit simulated (e.g., one driver, groups of drivers or vehicle fleets) and the number
of unit that can be simulated at once; (ii) visualisation, in terms of drivers' states,
parameters that can be observed in simulation and traffic; (iii) the ease in using
these models by other researchers.

2.2.2.1 In-Vehicle Information System - Jon Hankey

The purpose of in-vehicle information system (IVIS) is to provide a proof of
concept design tool in order to assess and compare in-vehicle system concepts,
prototypes and products for system designers and to provide insight into potential
design improvements. The approach for the development of this model consists of
compiling, analysing and using data from actual empirical research for creating
'micro-models' of performance prediction. The model predicts the visual and
cognitive resources needed at a basic level. These predictions are made at the
subtask level, because subtasks often require different resources and subtasks can
be combined into tasks which can themselves be combined into system assessment.
Therefore, the model accounts for a variety of information input, information
processing and response output combinations.
The overview of the IVIS demand model is shown in Fig. 2.2. During the
workshop, the emphasis was put on the software development of the tool and how
users would manipulate the tool. In terms of driver model, if we observe Fig. 2.2,
the component that most closely relates to the information processing is the one
listing driver resources involved, i.e., visual demand, auditory demand, manual
demand and speech demand. The assumption of the model is that the driver is a
finite capacity, with a single-channel processor of visual information. The driver
can 'share' cognitive resources to differing degrees and the resources needed to
30 Cody and Gordon

Task Specification in Terms of Driver Resources Involved

Visual Demand Manual Demand
Auditory Demand Speech Demand
Supplemental Information Processing Demand
I

Comparison/Interpolation Recall Task From

to Task(s) Found in Library Previously Saved Analysis

Program Identifies Task-Relevant Nominal Values for Primary Measures

Modification of Values, As Desired

Task Specific Subtask Specific

TaskllVIS Evaluation Using Figures of Demand

and Derived Values for Primary Measures

FIGURE 2.2. Overview of IVIS demand model.

perform IVIS tasks compete with the resources needed to drive the vehicle. In order
to avoid reducing the driving performance to unacceptable levels, only limited
resources can be required by IVIS. The different resource components and/or
magnitudes can be used to perform the same task and are required for different
tasks. The combination of the demand of these specific components determines the
required resources to perform a task. The required resource demand can be used to
estimate the potential of a decrement in driving performance. The principle of IVIS
is that nominal values for measures are derived that can be modified to match a
task or design specification; for example, a subtask modifier is the message length
and a task modifier is the roadway complexity.
The scale of simulation is one driver within one of the three age groups. The
software allows visualising the behaviour change outcome and the modifiers that
can be selected to adjust. Finally, the source code and a user manual are available
from FHWA. The software is still in a proof of concept stage. The model is used and
designed to be used in the industry, which can make the feedback to the scientific
community a slower process. For a review of the validation carried out at VTTI,
the reader can consult Jackson and Bhise (2002).

2.2.2.2 ACT-R Driver Model- Dario Salvucci

The theoretical endeavour of the ACT-R driver model is to provide a psycholog-
ically plausible model of an individual driver and combine cognitive, perceptual
 2. TRB Workshop on Driver Models 31

Declarative Module
(TemporallHippocampus

External World
FIGURE 2.3. ACT-R architecture.

and motor dimensions (Salvucci et al., 2001) . The approach involves applying
a cognitive architecture, ACT-R associated to a computational framework . The
advantage of this approach is the possibility to re-use theories and mechanisms
already integrated within the cognitive architecture. This approach currently fo-
cuses on highway driving involving moderate traffic. The two current practical
applications are (i) the prediction of driver distraction, for which models of typical
secondary tasks (e.g., phone dialling) are integrated into the model in order to pre-
dict real-world observables measures and (ii) the recognition of driver intentions,
where many models are run simultaneously, each trying to accomplish a different
goal and the method consist of tracking which model best matches the observed
data.
Figures 2.3 and 2.4 depict two diagrams: one for ACT-R and the othe r for ACT-
R driver model. Figure 2.3 shows (Anderson et a\., 2004) ACT-R architecture
and details the overlap between brain structures and information processing steps.
Figure 2.4 describes (Salvucci et a\., 2001) the component involved in the ACT-R
driver model. ACT-R provides a framework and specific rules are developed for
the driving model. The model also integrates cognitive models for other tasks, such
as using a cell phone or a navigation system. The current scale of simulation for
the ACT-R driver model is focused on one driver in an environment that provides
32 Cody and Gordon

With probability P1lU11dtor Iflead caris too close

FIGURE 2.4. ACT-R driver model representation.

interactions with other vehicles. This effort also includes the modelling of younger
versus older drivers. The visualisation allows observing driver's eye view and
mirror and in a new system currently under development, graphs will provide
measure of behaviour. Regarding the ease in using a model, the current version
of the ACT-R driver model cannot be used easily by other researchers, but a new
version is under development that should be easy to use (Salvucci et al., 2005).
In terms of calibration and validation, the approach applied for the ACT-R driver
model is the comparison of model and human data sets recorded similarly in terms
of real-world measures. The data sets that have been collected so far are highway
driving, phone dialling and distraction, radio tuning and distraction and the age
effects. Examples of measures are lane change steering profiles, gaze distribution
on highway, lateral deviation during phone dialling.

2.2.2.3 Optimal Control Model- Richard van der Horst

The driver modelling effort at TNO aims at evaluating driver behaviour, perfor-
mance and workload and at providing inputs for traffic flow model, such as MIXIC
driver model and human-kinetic traffic flow model. This approach focuses on indi-
vidual driver and vehicle units and is based on experimental research. The frame-
work applied for this modelling effort is the one of 'optimal control', i.e., the use
of a linear system theory, where the assumption is that the driver is well trained and
well motivated to behave optimally. The model also integrates inherent limitations
and constraints. The resulting model is a realistic description of driver behaviour
in terms of driver performance, workload and total system performance measures.
 B-1
2. TRB Workshop on Driver Models 33

DRMR H VEHICLE ~ BEHAVJOR

f
FI GURE 2.5. Block diagram for driver/vehicle/system and driver/model.

Two figures support the model architecture. Figure 2.5 is a block diagram of
the driver vehicle system. In this figure , the concept of task covers elements such
as lane keeping , car following or speed control , while the behaviour is described
as measures of performance and workload . Figure 2.6 illustrates a 3 x 3 x 3
description of driver behaviour (Theeuwes, 2001) relative to the task hierarchy,
task performance and information processing . The shaded area denotes the part
of the behaviour that is currently modelled , i.e., the control of the vehicle at a
skilled-based level. For this part of the behaviour, all of the information proce ss
is integrated, from percept ion to action. Figure 2.6 provoked a lot of interest from
the audience, as it is commonly considered that strategic equal s knowled ge-based
level, manoeu vring equal s rule-b ased level and control equals skill-based level. A
question was raised about how to transition within this 3 x 3 x 3 representation.
For example , how can a driver be simulated at the skill-based control and then
simulated at the knowledge-based level? Is it only by learning? Can it also be
due to other factors? The answer pointed to situations such as degraded driving
co nditions. For exampl e, an experienced driver mobili ses skills (Task performance)
at the control level (Task hierarchy), but while driving another vehicle will move
the vehicle control one level up to the rule level. Another example given was the
case of a novice driver, for whom the control of the vehicle can be associated to a
kn owledge-level perform ance.

>-
/
:I: ,/
U ,/ / v
0:::
<
0:::
strateg ic ///

,./.
W maneuvering A /l.-(
:I:
~ control

s
In

FI GURE 2.6. Driver model.

34 Cody and Gordon

World
IObjects I 4--------------------------------------------------------
vehicle dynam ics

---------1

r
Driver ~ - - -- , , --- - -- ., ,... ---
Sensors
I
Processor I
Body
Eyes, 1 s ensor-r H Action- Anns , Legs ,
Ears. filters filter Head
- - _... , I, , -- - -
Proprioception
---- ---,
I
.a.
,
I I ..., 0

---- ----- ---- ----------,

0
- - -- - ----- - -- - ------- -- ~
attent ion control haptic feedback

Vehicle
--------- --- ----- ------ --- ---- ---- ----- ----- --- -- - Controls
Field of view Pedals ---
Steering whee l.. .

FIG URE 2.7. ACME driver model.

The outputs generated by the OCM are time simulation results , performance
and workload measures and a workload index. The example of validation provided
during the workshop was a comparison of data collected on a driving simulator and
of model prediction on the lane-keeping task . The agreement between the model
and experimental results indicates a useful predictive capability.

2.2.2.4 ACME
The purpose of the ACME driver model it to develop a combined view on car-
following and lane-changing as well as describe and evaluate the mental processes
needed for driving . It is a man in the loop simulation of a 'car-driver' unit composed
of three sub-models: model of human sensors, model for information processing
and a model for action execution. The simulation integrates models of the dynamics
of the vehicle and models of the simulated area.
The ACME driver model is a very modular architecture (Fig. 2.7) and one of the
most driver vehicle infrastructure system oriented among the presentations given.
The model can be divided into three major substructures: (i) senses, (ii) informa-
tion processing and (iii) actions execution. The senses that are implemented are
vision and hearing, and the perception of acceleration is used to influence speed
decision and the haptic input is not integrated. The information processor consists
of an internal world representation storing , a planning instance and an execution
instance. In order to carry out the execution of action, the extremities are simulated
to move in-vehicles devices to certain position. These devices in tum determine the
vehicle's dynamism. Regarding the scale of simulation, approximately 20 vehicles
can be simulated around an intersection. The simulation step ranges from 10 to
100 ms. The visualisation represents the states within the driver's cognition and
includes timelines of measures. Regarding the ease of using the model, it would
 2. TRB Workshop on Driver Models 35

While driving, you are...

Watc hing

In Fuzz y Logic, this becomes

Fuzzy Input Fuzzy Firing Defuzzification

x (relative speed) =A' If x =A and y = B then z = C Because x =A' and y = B'

y (headway divergence) = B' z (acceleration rate) = C'

FIGURE 2.8. FLOWSIM driver model.

still require time for the user to become familiar with the model and its implemen-
tation. The ACME model is still in the process of development and hence has not
been validated yet.

2.2.2.5 Fuzzy Logic Based Motorway Simulation

The development of Fuzzy LOgic based motorWay SIMulation (FLOWSIM)

started in 1997. The model was initially focused on highway traffic, but now covers
all type of roadway. This framework was first used for advanced driver assistance
system (ADAS) and infrastructure speed control applications. Recent urban appli-
cations include network travel time prediction. It is now the object of an intensive
enhancement program in China and has been chosen as the traffic simulation tool
for supporting the city of Tianjin's transportation planning and management.
FLOWSIM driver model (Brackstone, 2000; ;rackstone et aI., 1997 ; Wu et aI.,
2000) representation (Fig . 2.8) presents the three basic steps of watching, thinking
and responding and how these basic steps can be associated to fuz zy logic. The
example used for the description of the model focused on speed and gap control. It
showed how the relative speed and distance divergence influence the driver 's action
in terms of acceleration rate . This model allows most units to be simulated during
one simulation, with the possibility of simulating up to a 1000 vehicles. Different
behaviours are simulated by using distributions . The speaker found little empirical
evidence about the existence of groups such as young or old or aggressive drivers
and therefore uses the distribution without labels tying the driver/vehicle unit to
specific groups. The visualisation mainly displays individual vehicles and the road
geometries. In order to use the system, it would be necessary to receive training.
The validation of FLOWSIM (Wu et aI., 2003) concentrated on the comparison
of simulated and measured traffic flows and showed a very satisfying fit.
36 Cody and Gordon

2.3 Synthesis of Presented Models

This section presents two sub-sections. In the first one, we introduce a set of
dimension and visualisation methods to understand the scopes of the different
models that were presented above, their commonalities as well as their dif-
ferences. In the second section, we will discuss briefly the notion of driver
model as a tool used for the development or evaluation by users who are not
necessarily human sciences scientists and, although they need to manipulate
driver model, should not have to create all of the elements going into a driver
model.

2.3.1 Understanding Models' Scope

We consider that the basic architecture of a simulation involving a driver
model is constituted of at least three main components: (i) a driver model,
(ii) a vehicle model and (iii) an infrastructure model. This hierarchy (Fig. 2.9)
can be further expanded into increasing the levels of detail. For example, the
driver model can be expanded in more modules or dimensions, such as driv-
ing tasks classification (strategic, tactical and operational) or psycho-motor
dimension, which is constituted of perception, cognition and motor control.
For the workshop, we defined a number of dimensions for each of these
three main components and asked the speakers to rate their models on these
dimensions.

Driving tasks Vision

Driver model perception

Psycho-motor Auditory
cognition

Motor control
Vehicle response
Vehicle(s) model

Vehicle motions

level of details

Infrastructure model

Type

FIGURE 2.9. Hierarchy of model components.

 2. TRB Workshop on Driver Models 37

For the driver component, the proposed dimensions were:

psycho-motor
driving tasks
control level
model capacities
simulated phenomena
simulation control

For the vehicle component, the proposed dimensions were

control variables
vehicle motions
vehicle subsystems
vehicle response

And, for the infrastructure component, the two main proposed categories were

level of details (lane, signalisation)

environment type (highway, urban ....)

As the workshop was limited in time, we focused the scope of the description to
the three main components and on the details of the psycho-motor dimensions.
The rating was as follows:
0: not represented
1: indirectly represented (there is a related parameter)
2: basic inclusion (there is a specific parameter in the model that indicates relevant
trends)
3: included (the parameter or state is directly represented via a simple sub-model)
4: modelled (a sub-model represents the process)
5: represented in detail (the process is a core aspect of the model and be related in
some detail to experiments or theories)
This rating was then integrated in a 'radar web' graph. Each speaker presented
a graph for each of the three components and then a more detailed graph of the
psycho-motor level. The ratings for all the models were then integrated on a same
graph (see Figs. 2.10 and 2.11).
The purpose of representing the models presented at the workshop via a common
graph was to identify synergies and limitations. In order to categorise synergistic
dimensions, it was proposed that when a dimension was ranked three or above
by at least three models, then a synergy between the three models is possible. As
shown in Fig. 2.10, the dimensions that fit this category are perception, cognition,
operational, rule, reaction and static input. A second category is one where di-
mensions could clearly be further developed, where at least one model ranks three
or above. The dimensions in this category are motor control, strategic, tactical,
skill, non-driving activities, knowledge, anticipation, distraction and driver char-
acteristics. Another interesting category is one where none of the models ranked
 38 Codyand Gordon

Perception Ps cho-motor

Simulation
Control

Driving
driver characteristics tactical Tasks

reaction time~~~~~~~~~~~~~l;;;9~~~Joperational
Simulated

Control Levels
Model Capacities
non driving activities

IVIS ..... ACT-R ..... Workload - ACME ~ Flowsim I

FIG URE 2.10. Representation of drivermodel dimensions.

three or above with the following three dimensions: learning , variable and dynamic
inputs.
The same method was applied for the description of the psycho-motor dimen-
sions (Fig. 2.11), and the areas for which we identified synergies are vision, decision
making , memories, steering control and velocity control. The areas that need to
be developed are haptic, auditory, recognition, anticipation, rules/knowledge and
attentional resources. Finally, the areas still to be covered are tactile and other
vehicle control.
This categorization of the dimensions led to a question about researchers imple-
mentation strategy and what dictates the choice of what to implement first. Are the
dimensions from the first category receiving so much attention because they are
easier to implement than the low ranking ones or because they are more important
to the concept of driver model and its current applications?

2.3.2 Driver Model Toolbox

The notion of a driver model toolbox came up during the workshop discussion. A
parallel was drawn with tools such as Matlab and Simulink, which allow the use
of pre-existing tools to create simulations; for driver models, researchers have yet
to start developing such a simulation environment. The two main aspects covered
were how the toolbox should be developed and how to identify potential user
group s.
 2. TRB Workshop on DriverModels 39

Vision
Perception

Motor Control

steering contro' 1\~f-l-:l-~~SJ~ ~=7==;;P'i- l

attentional resources

Cognition
I...... IVIS " ' - ACT-R ..... Workload ACME .... Flowsim I

FIGURE 2.11. Representation of psycho-motor dimensions.

Matlab is a commercial software package supporting simulations from most en-

gineering disciplines, and it is difficult to aim at a completely similar product. The
equivalent would be a commercial software supporting simulation of intelligent
systems or of the human mind . One of the open suggestions was to create a pub-
licly available tool, based on the contributions of an international consortium of re-
searchers. Such a tool would be a ' driver simulator' software, which would allow to
observe as a simulation is conducted the different components of the model activity .
The two key groups - yet to be identified - are a scientist/researcher group,
who would design and develop the toolbox, and a parallel user group, who would
test, develop and help validate the resulting models . It is anticipated that whilst
these two groups would have a range of differing requirements and interests,
their activities could be complementary and lead to a new general level of driver
modelling capability.
In some respect, the models closest to achieving this role are the IVIS and
ACT-R driver model. The limitations of these models are their limited scope , on-
going calibration and validation and for the ACT-R driver model, the step of going
towards a more user friendly platform and interface.

2.4 Towards a Comprehensive Model of Driving

Any comprehensive model of driving should be capable of reproducing statisti-
cally verifiable trends in response to changes made to its constitutive parameters,
40 Cody and Gordon

and those parameters should be rich enough to encompass the interests of a very
wide class of transportation researchers - across the disciplines mentioned in the
Introduction (Section 2.1). Roughly speaking, relevant time-based responses (eye
glance, steering, emergency braking, decision latency, etc.) should be available to
predict trends as key parameters of interest change (external vehicle behaviour,
vehicle control system activity, driver experience, vehicle information and warning
systems, driver-vehicle interface, distracting activities, external signalling, etc.).
No such model exists at present; indeed, it is probably naive to expect that any
single model will ever be sufficiently full and complete to represent all possible
behavioural changes in response to all possible parametric changes. However it is
plausible to expect that a single modular formulation of the driving process - or
more precisely a single functional representation of driving - could be developed
to accelerate the progress on any particular model-based question of this type.
A common modular framework is also an essential starting point for any se-
rious coordinated effort in driver modelling. In the foregoing text (Fig. 2.9) we
have suggested the general scope of such a modular approach, but the framework
itself is undefined. The aim here is to start to define what such a framework might
look like, without necessarily proposing this in any final form. Figure 2.12 illus-
trates the point. Each rectangular shape depicts a process that is at the same time
stochastic and predictable. Those associated with the driver are open to learning
and adaptation. Note that in this representation, the driver appears as a distributed
set of processes! Indeed, the structure embodies a whole range of assumptions and
hypotheses. For example, it suggests that some form of 'vision-based' driving is
possible without the involvement of higher cognitive function, but that manoeu-
vring decisions are not possible without (at least occasional) strategic input. The
diagram is not intended to be complete (e.g., there is no auditory information chan-
nel) or even correct; indeed the correctness of this or other functional maps of its
kind ultimately depends on its ability to match experiments.

2.5 Conclusions
The models presented at the workshop varied based on the goal for which they
are designed and the methods applied to implement, calibrate and validate them.
However, they do share commonalities in the processes that they manipulate. The
authors' intention when convening the speakers was to illustrate the variety of
driver models and to convey the point that the aim of developing driver models is
not to create 'the' driver model or a driver model representing the right approach.
Actually, the level of detail to integrate in a driver model really depends on the goal
of the developer. For instance, in order to predict driver behaviour, how necessary
is it to describe the psycho-motor processes underlying the driving activity or is a
data analysis of data patterns and trends sufficient?
The intent of the workshop was to open up the discussion on modelling issues
and exchange, move from a 'researcher centric' to a community-directed approach
in order to take driver model development further. The other goal was to discuss the
 2. TRB Workshop on Driver Models 41

strategic

DVI

vehicle vehicle
control
dynamics kinematics

:=
+-- other
vision vehicles

infrastructure

FIGURE 2.12. Possible functional representation of the driving process.

idea of extending the notion of driver model towards the concept of comprehensive
driving model, where the driver becomes one the component of a system and could
become a tool used in order to support ITS development or other safety application
for which it is necessary to know more about the driver behaviour or information-
processing characteristics. In this sense, this concept differs from classical traffic
simulation tool by the scale at which it is envisaged, although simplified versions
could eventually be coupled with more conventional traffic simulation tools. In this
regard, the aim is to take the concept of driver model out of the research community
and bring it to a wider range of users, who can be engineers designing in-vehicle
systems and needing to understand the impact of their systems on driving or traffic
engineers needing to use driver models when changing a roadway design without
systematically having to conduct lengthy data collection.
The next step to pursue the development of a comprehensive driving model will
be to continue to take advantage of conferences to bring together model developers
and users in order to discuss about the elements to include on a driving model and
methods for exchanging the results of their research.
42 Cody and Gordon

Acknowledgments. We acknowledge the benefits we had from our exchanges with

the researchers who participated in the workshop. We also give thanks for the many
fruitful discussions we had with Jon Hankey, Dario Salvucci, Daniel Krajzewicz,
Mark Brackstone and Richard van der Horst during the preparation of the workshop
and with the audience on the day of the workshop.

References
Anderson, 1.R., Bothell, D., Byrne, M.D., Douglass, S., Lebiere, C. and Qin, Y. (2004) An
integrated theory of the mind. Psychological Review, 111, 1036-1060.
Brackstone, M. (2000). An examination of the use of fuzzy sets to describe relative speed.
Perception Ergonomics, 43(4), 528-42.
Brackstone, M., McDonald, M. and Wu, 1. (1997). Development of a fuzzy logic based
microscopic motorway simulation model. In Proceedings of the IEEE Conference on
Intelligent Transportation Systems (lTSC97). Boston, MA.
Jackson, D.L. and Bhise, V.D. (2002). An evaluation of the IVIS-DEMAnD driver attention
demand model. Report No. SAE 2002-01-0092, UMTRI-95608 A08. Human Factors in
Seating and Automotive Telematics, Warrendale, SAE, pp. 61-70.
Salvucci, D.D., Boer, E.R., and Liu, A. (2001). Toward an integrated model of driver
behavior in a cognitive architecture. Transportation Research Record, 1779, 9-16.
Salvucci, D.D., Zuber, M., Beregovaia, E., and Markley, D. (2005). Distract-R: Rapid
prototyping and evaluation of in-vehicle interfaces. In Proceedings ofHuman Factors in
Computing Systems. Portland, OR.
Theeuwes, 1. (2001). The effects of road design on driving. In Pierre-Emmanuel Barjonet
(Ed.). Traffic Psychology Today (pp. 241-263). Kluwer, Boston, MA.
Wu, 1., Brackstone, M. and McDonald, M. (2000). Fuzzy sets and systems for a motorway
microscopic simulation model. Fuzzy Sets and Systems 116(1), 65-76.
Wu, 1., Brackstone, M. and McDonald, M. (2003). The validation of a microscopic simula-
tion model: A methodological case study. Transportation Research C, 11(6),463-479.
3
Towards Monitoring and Modelling
for Situation-Adaptive Driver
Assist Systems
TOSHIYUKIINAGAKI

3.1 Introduction
In the classic tri-level study of the causes of traffic accidents, Treat et al. (1979) as-
cribe 92.6% of car accidents to human error, where human errors include improper
lookout (known as 'looking but not seeing'), inattention, internal distraction and
external distraction. Green (2003) reports that other studies have found similar
results that a human error is involved in 90% of car accidents. Human errors, such
as those listed in the above, can happen for everybody and may not be eradicated.
However, if there were some technology to detect driver's possibly risky behaviour
or state in a real-time manner, car accidents may be reduced effectively. Proactive
safety technology that detects driver's non-normative behaviour or state and pro-
vides the driver with appropriate support functions plays a key role in automotive
safety improvement. Various research projects have been conducted worldwide to
develop such technologies (see, e.g. Witt, 2003; Panou et al., 2005; Saad, 2005;
Amditis et al., 2005; Tango and Montanari, 2005; Cacciabue and Hollnagel, 2005).
This paper gives an overview on two of research projects in Japan, which aim
to develop technologies to detect driver's behaviour or state that is inappropriate
to a given traffic environment so that the driver may be provided with support for
enhancing his or her situation awareness or for reducing risk in the environment.
The first project is the 'Behaviour-Based Human Environment Creation Technol-
ogy', which was conducted during the period of 1999 to 2003, with the support of
the Ministry of Economy, Trade and Industry (METI), Government of Japan, and
the New Energy and Industrial Technology Development Organization (NEDO).
The other project is the' Situation and Intention Recognition for Risk Finding and
Avoidance' which has been proceeding since 2004, with the support of the Ministry
of Education, Culture, Sports, Science and Technology (MEXT), Government of
Japan.
This paper tries to focus on the modelling-related aspects of the projects, and
picks up a model from each project. From the first project, is the Bayesian network
model for detecting non-normative behaviour of the driver, in which the model
has been constructed based on driving behavioural data collected in the real traffic
environment. From the second project, is the discrete-event model of dynamical

43
44 Inagaki

changes of driver's psychological state, which has been developed to analyse and
determine how decision authority should be distributed between the driver and
automation under possibility of the driver's overtrust in 'smart and reliable' au-
tomation. The models are still in their early stages; however, they are expanding
description capabilities and applicabilities in the real world.

3.2 Behaviour-Based Human Environment Creation

Technology Project
3.2.1 Aims of the Project
When a driver's performance is suitable to task demands determined by traffic
situations (such as vehicle performance, vehicle speed, road structure, weather,
other traffic), the driver can enjoy safe driving. However, if the driver's performance
fails to adapt to the task demands, his or her behaviour may increase risk to the
situation (see Fig. 3.1). In order to evaluate the level of risk, traffic situations as
well as driving performance need to be sensed and monitored.
The Behaviour-Based Human Environment Creation Technology project was
conducted during the period of 1999 to 2003, with the support ofMETI and NEDO .
The project had four aspects, and one of the aspects was to create behaviour-
based driving assist technologies that determine whether a driver's performance
is deviated from the normative one in a given traffic situation as well as provide
the driver with a warning or advice, when appropriate. A driver's normative (or
baseline) behaviour was defined as his or her usual behaviour. The project took as an
object of study a driver who usually bears safe driving in mind and used SAS 592, a
self-rating scale of driving attitude , developed by the National Research Institute of
Police Science, Japan, to exclude inherently unsafe drivers from the data collection
phase .

Crash
(Fail to perfonn task)

FiGURE 3.1. Driving performance and task demand (Akamatsu and Sakaguchi , 2003); orig-
inally from Fuller and Santos (2002) .
3. Towards Monitoring and Modelling for Situation-Adaptive Driver Assist Systems 45

3.2.2 Measurement ofDriving Behaviour

In order to detect a driver's deviation from his or her normative performance,
good knowledge on the driver's usual performance is necessary. Vehicles with
a driving recorder system were developed, for the project, to record and collect
behavioural data in real road environments. The system consists of sensors, small
CCD cameras, a signal-processing device and a laptop on-board computer. The
recorded data include steering wheel angle, tum signal, wiper activation, strokes
of the brake and gas pedals, position of the right foot of the driver, geographical
position of the vehicle, velocity and state of the vehicle, and relative distance and
speed to the lead and following vehicles.
Eight different driving routes were chosen for investigation as 30-min trips
with several right and left turns. Subjects were recruited and asked to drive a
specified route once a day on weekdays for 2 months. Forty sets of behavioural
data were collected for each subject per route (Akamatsu et al., 2003). At the end
of data-collection phase, each subject was given the Driving Style Questionnaire
developed for quantifying driver's attitudes, that may be related to or may affect
his or her driving behaviour (Akamatsu, 2003). The total number of participants
was 92 (59 male and 33 female). The age ranged from 21 to 71 years for males
and from 20 to 66 years for females.

3.2.3 Driving Behaviour Modelling

A Bayesian network approach has been applied to driving behaviour modelling.
Bayesian networks (often called belief networks) are directed graph models, in
which an arc from a node, say A, to another, say B, represents a causal relation-
'A causes B'. The conditional probability distribution is specified at each node in
the network to represent how a child node is affected by its parent nodes.
For each situation of interest, a Bayesian network model can be developed.
Consider a case, for instance, in which a driver approaches an intersection with the
STOP sign. Behavioural events needed for describing the case include (a) release
of the gas pedal, (b) moving the foot to the brake pedal, (c) onset of braking,
(d) reaching the maximum deceleration and (e) full stop. Figure 3.2 depicts the
resulting Bayesian network model, obtained based on the collected behavioural
data. In the Bayesian network modelling, the data analysis results used were those
claiming that the weather conditions and the driver's score for the methodical scale
in the Driving Style Questionnaire can be regarded as performance-shaping factors
(Akamatsu and Sakaguchi, 2003).

3.2.4 Detection ofNon-Normative Behaviour

Once a Bayesian network model has been obtained for a driver's normative be-
haviour, it then becomes possible to compute the probability distribution for each
node in the network. Consider a case, as an example, in which some node, say
Z, denotes a random variable related to a safety-critical action for avoiding an
accident. Suppose Z has two nodes, X and Y, as its parents. Based on probability
46 Inagaki

FIG URE 3.2. Bayesian network model for the behavioural events when approaching an
intersection with the STOP sign (Akamatsu and Sakaguchi, 2003).

distributions for X and Y and the knowledge of probabilistic causal relation among
the nodes, one can derive the conditional probability distribution, P(ZIX , Y) for
each combination of values of X and Y. The conditional probability distribution
P(ZIX , Y) describes how Z can take different values when the driver behaves in
a normative manner.
Suppose an on-board sensor has observed that the random variable Z took a
value z at some time point. Hypothesis testing is then performed to determine
whether 'z may be regarded as a sample from the distribution P(Z IX, Y)'. If the
hypothesis was rejected at some level of significance, it is then concluded that
the driver's behaviour may be deviated from his or her normative performance.
Akamatsu et al. (2003) defined the ' level of normality' and have developed a
method to give a warning to the driver when the calculated level of normality
becomes less than a specified threshold value .

3.2.5 Estimation ofDriver's State

In the Behaviour-Based Human Environment Creation Technology project, meth-
ods for estimating driver's mental tension and fatigue were also investigated, be-
cause driver's status need s to be estimated to provide a driver with support functions
suitable to his or her operational capability.

3.2.5.1 Estimation of Driver 's Mental Tension

Mental tension is one of driver's internal factors affecting driving ability. Investiga-
tors in the project have found that chromogranin A (CgA) in saliva is a biochemical
3. Towards Monitoring and Modelling for Situation-Adaptive Driver Assist Systems 47

index of driver's mental strain (Sakakibara and Taguchi, 2003). The index was not
convenient for real-time sensing of driver's mental tension. They found, however,
that the steering operation and the head motion of a driver can replace salivary CgA
in estimating a level of driver's mental tension. Taguchi and his colleagues argued
that the steering operations at a frequency lower than 0.5 Hz may reflect a driver's
reduced activation state in a monotonous driving environment and that the driver's
intentional behaviour, such as a lane change, does not usually have influence on
operational performance at such low frequencies (Sakakibara and Taguchi, 2005;
Taguchi and Sakakibara, 2005). They also considered that while concentrating
on driving, a driver generally puts power into various muscles to control body
in response to vibrations from road surface, acceleration and/or deceleration of
vehicle, which may create stiffness in the shoulders. Experiments were conducted
to investigate relations between the two indices for tension. It has been observed
that the steering operation and the head motions are complementary to each other.
They have also found that a real-time estimation method can be implemented to
distinguish the four grades of driver's tension (viz. reduced activation, neutral,
moderate tension and hypertension).

3.2.5.2 Estimation of Driver's Fatigue

Fatigue is also one of the contributing factors affecting driving ability. Investigators
in the project have tried to develop a non-invasive sensing method for real-time
estimation of a driver's fatigue level (Furugori et al., 2003, 2005; Miura et al.,
2002). With sensor sheets, Furugori and his colleagues measured the pressure
distribution on the seat and found a relationship between changes in the load
centre position (LCP) calculated from the pressure distribution and driver-indicated
subjective fatigue. It was also found that the LCP contains information about
'prolonged changes' in posture with time (e.g. bending forward or backward) as
well as 'momentary changes' in posture (e.g. shifting weight). A 'postural change
parameter' and a 'weight-shifting parameter' were defined for the seatback and
the seat cushion. However, individual differences were observed in direction of
postural changes and weight shifting. A robust algorithm was thus required to
estimate fatigue level for each individual. Based on the ratios that indicate the
degree to which each parameter range deviates from those in normal driving, the
'fatigue index' was calculated to estimate a driver's fatigue level. It was found that
in around 90% cases the calculated fatigue index agreed with subjective fatigue
(Furugori et al., 2005).

3.3 Situation and Intention Recognition for Risk Finding

and Avoidance Project
3.3.1 Aims of the Project
Driving requires a continuous process of perception, decision and action. Under-
standing of the current situation determines what action needs to be done (Hollnagel
48 Inagaki

traffic environment

I perception I0::::::::::::::::;> L...---r------J c==;=~:.L.~~J

infer "intent"

"Is the driver's interpretation appropriate?"

Detect: mismatch ofbehavior and environment

FiGURE 3.3. Assessing appropriateness of a driver's situation recognition.

and Bye, 2000). In reality, however, drivers' situation recognition may not always
be perfect. Decisions and actions that follow poor or imperfect situation recogni-
tion can never be appropriate to given situations. It is not possible to 'see' directly
whether a driver's situation recognition is correct or not. However, monitoring the
driver 's action (or its precursor) and traffic environment may make it be possible
to estimate (a) whether the driver may have lost situation awarenes s, (b) whether
the driver's interpretation of the traffic environment is appropriate and (c) whether
the driver is inactive psychologically, e.g. due to complacency, or physiologically,
e.g . due to fatigue, (see Fig. 3.3).
Since 2005, the author has been conducting a 3-year research project , enti-
tled ' Situation and Intention Recognition for Risk Finding and Avoidance', with
the support of the MEXT. The aim of the project is to develop proactive safety
technologies to (a) monitor driver behaviour for assessing his or her intention, (b)
detect mismatches of traffic environment and driver's interpretation of it, (c) assess
the driver state and (d) provide the driver with appropriate assist functions in a
situation-adaptive and context-dependent manner (see Fig. 3.4).
The research topics in the project are categorised as follows : (1) estimation of
driver 's psychological and physiological state, (2) driver behaviour modelling, (3)
intelligent information processing for situation recognition and visual enhance-
ment and (4) adaptive function allocation between drivers and automation. In (I),
real-time methods are under development for detecting the driver's inattentive-
ness , hypo-vigilance, and complacency, through monitoring parameters, such as
body movement, dynamical changes of the LCP on the back, eye and head move-
ments , blinks , instep position of the right leg, operation of the steering wheel
and movements of gas and brake pedals (Itoh et al., 2006) . In (1), levels of
 w
~~ .:_ =-~ ~ ~. ,
'.--:--. . . -::4-- _.., ~ -. Q3
:E
~
- ''' , . .,
- -. '"a.
'"
~
o
::l
s"
:::!.
Driver's behaviors ::l
oe
::l
'"0-
I ~
o
0-
~
s
{Jt>

..,0'
co
;:
;:;
o
::l
:i>
0-
-g
'"
<.
rt>

.o.,
Risk finding <.
rt>
..,
:>
'"
'"
~"
(/)
'<
o
n;
Human -centered accident-prevention technology for transportation safety 3
'"
.j:>.
FIGURE 3.4. Situation and intention recognition for risk finding and avoidance project. \Q
50 Inagaki

driver's fatigue and drowsiness are also estimated by applying a chaos theoretic
method (Shiomi and Hirose, 2000) to a driver's voice during verbal communi-
cation. Driver modelling in (2) adopts a Bayesian network approach, as in the
case of the Behaviour-Based Human Environment Creation Technology project.
Some mathematical and information processing methods are under development
in (3) for machine learning, recognition of traffic environments and human vision
enhancement.
The methods in (1) to (3) give messages (or warnings) to the driver, when
they determine that the driver's situation recognition and intention may not be
suitable to a given traffic condition. If the driver responds quickly to the messages,
the potential risk shall be diminished successfully. If the driver fails to accept
or respond to the messages in a timely manner, on the other hand, accidental or
incidental risks may grow. Research topics in (4) investigate such situations. They
aim to develop an adaptive automation that can support drivers at various levels of
automation, which shall be discussed in the next section.

3.3.2 Adaptive Function Allocation Between Drivers

and Automation
A scheme that modifies function allocation between human and machine dynam-
ically depending on situations is called an adaptive function allocation. The au-
tomation that operates under an adaptive function allocation is called adaptive
automation (see e.g. Inagaki, 2003; Scerbo, 1996). Adaptive automation assumes
criteria to determine whether functions have to be reallocated, how and when.
The criteria reflect various factors, such as changes in the operating environment,
loads or demands to the operators and performance of the operators. Adaptive au-
tomation is expected to improve comfort and safety of human-machine systems in
transportation. However, it is known that the humans working with highly intelli-
gent and autonomous machines often suffer negative consequences of automation,
such as the out-of-the-loop familiarity problem, loss of situation awareness, au-
tomation surprises. When carelessly designed, adaptive automation may face with
such undesirable consequences. One of critical design issues in adaptive automa-
tion is decision authority over automation invocation (viz. who makes decisions
concerning when and how function allocation must be altered). The decision au-
thority issue can be discussed in a domain-dependent manner. Automobile is one
of domains in which machine-initiated control over automation invocation may be
allowed for assuring safety (Inagaki, 2006). Following are two examples that are
under investigation in the project.
Example 1: Suppose that the driver of the host vehicle H determines to make a
lane change because the lead vehicle A has been driving rather slowly. When the
host vehicle's on-board computer noticed that the driver glanced the side mirror
several times, it inferred that the driver had formed an intention of changing lane,
where the computer regarded 'glancing the side mirror several times in a short
period of time' as precursor to the action of changing lane (see Fig. 3.3). The
computer, monitoring backward with a camera, also noticed that a faster vehicle
3.Towards Monitoring and Modelling for Situation-Adaptive Driver AssistSystems 51

A H

B c D

FIGURE 3.5. An examplein which machine intelligence is given decision authority.

C is coming from behind on the left lane . Based on the understanding of driver's
intention and the approach of vehicle C, the computer puts its safety control
function into its armed position in preparation for a case when the driver chooses
a wrong timing to execute an action (viz. steering the wheel) due to improper
interpretation of the traffic environment. Now the driver, who has seen that a
very fast vehicle B almost passed him on the left, begins to steer the wheel to the
left, failing to notice vehicle C (Fig. 3.5). The computer immediately activates
the safety control function to make the wheel either slightly heavy to steer (soft
protection) or extremely heavy to steer (hard protection). The soft protection is
for correcting the driver 's interpretation of the traffic environment, and the hard
protection is for preventing a collision from occurring. The computer takes the
steering authority from the driver partially in cases of soft protection and fully
in cases of hard protection.
Example 2: Suppose that the driver of the host vehicle H wants to make a lane
change to the left, because the lead vehicle A drives rather slowly. When glancing
at the rear view mirror, the driver noticed that faster vehicles, C and D, are
coming from behind on the left lane (Fig . 3.5). By taking several looks at the
side mirror, the driver tries to find a precise timing to cut in. In the meantime,
based on the observation that the driver looked away many times in a short
period of time , the on-board computer determined that the driver has formed an
intention of changing lane and that he might not be able to pay full attention
to the lead vehicle A. The computer then puts its safety control function into
its armed position in preparation for a deceleration of the lead vehicle A. If the
lead vehicle A did not make any deceleration before the host vehicle's driver
completes a lane change, the computer will never activate the safety control
function and will put it back into a normal standby position. On the other hand,
if the computer detected a rapid deceleration of the lead vehicle A while the
driver is still looking for a timing to make a lane change, it immediately activates
its safety control function, such as an automatic emergency brake .

3.3.3 Decision Authority and the Levels ofAutomation

For the discussion of decision authority, the notion of the level ofautomation (LOA)
is useful. Table 3.1 gives an expanded version in which an LOA comes between
levels 6 and 7 in the original list by Sheridan (1992). The added level, called
52 Inagaki

TABLE 3.1. Scales of levels of automation.

1. The computer offers no assistance; human must do it all.
2. The computer offers a complete set of action alternatives and
3. narrows the selection down to a few, or
4. suggests one and
5. executes that suggestion if the human approves, or
6. allows the human a restricted time to veto before automatic execution, or
6.5 executes automatically upon telling the human what it is going to do, or
7. executes automatically, then necessarily informs humans,
8. informs him after execution only if he asks,
9. informs him after execution if it, the computer, decides to.
10. The computer decides everything and acts autonomously, ignoring the human.

After Sheridan (1992), Inagaki et al. (1998) and Inagaki and Furukawa (2004).

level 6.5, was first introduced in Inagaki et al. (1998) with twofold objectives:
(1) to avoid automation surprises possibly induced by automatic actions and (2)
to implement actions indispensable to assure systems safety in emergency. When
the LOA is positioned at level 6 or higher, the human may not be in command.
Generally speaking, it would be desirable, philosophically and practically, that
human is maintained as the final authority over the automation. However, as can
be seen in Examples 1 and 2, there are cases in which automation may be given
decision authority (Inagaki, 2006).

3.3.4 Model-Based Evaluation ofLevels ofAutomation

This section gives a driver model for evaluating design of interactions between a
driver and automation under the possibility of the driver's overtrust in the automa-
tion. Let us take a case, as an example, in which an adaptive cruise control (ACC)
system is available on the host vehicle. The ACC system is a partial automation for
longitudinal control, designed to reduce the driver's workload by freeing the driver
from frequent acceleration and deceleration. It controls the host vehicle so that it
can follow a vehicle ahead (the target vehicle) at a specified distance. When the
ACC system detects the deceleration of the target vehicle, it slows down the host
vehicle at some deceleration rate. As long as the deceleration of the target vehicle
stays within a certain range, the ACC system can control the speed of the host
vehicle perfectly and no rear-end collision into the target vehicle occurs. It would
be natural for the driver to trust in the ACC system while observing it behaves
correctly and appropriately. Sometimes the driver may place excessive trust in the
automation and may fail to allocate his or her attention to the traffic environment.
If the target vehicle makes a rapid deceleration at a high rate, the ordinary brake
by the ACC system may not be powerful enough to avoid a collision into the target
vehicle. In such cases, the automation issues an 'emergency-braking alert', which
tells the driver to hit the brake himself or herself hard enough to avoid a collision.
If the driver has been in a hypovigilant state, however, he or she may fail to respond
quickly to the situation.
3.Towards Monitoring and Modelling for Situation-Adaptive Driver AssistSystems 53

t
State IV: Hyper-normal and excited state. Target vehicle ~
approaches

C~:::>
State III: Normal and vigilant state
(the best state for safe driving).
emerge.ncy
-brakmg
State II+: Normal and relaxed state with
moderate level of trust in the
t , ! alert
automation.
~ CJL:)
State II : Normal and relaxed state, with experiencing
complete faith in the automation.

State I: Subnormal and inactive state.

perfect control
consecutively CD
FIGURE 3.6. Drivers' psychological state and their dynamic transitions.

3.3.4.1 Drivers' Psychological States and Their Transitions

Inagaki and Furukawa (2004) have distinguished five psychological states for
drivers, by modifying the original model by Hashimoto (1984):
State I: Subnormal and inactive state.
State II: Normal and relaxed state, with complete faith in the automation.
State 11+: Normal and relaxed state with moderate level of trust in the automation.
State III: Normal and vigilant state, which is the best state for safe driving.
State IV: Hyper-normal and excited state.
State transitions occur dynamically in time as time passes by. Suppose a driver's
psychological state was positioned initially at State III, when he or she started
driving . If the driver has observed a certain number of the ACC system's perfect
longitudinal controls in response to decelerations of the lead vehicle , the driver's
psychological state goes from III to II+. Observing some more consecutive perfect
controls by the ACC system , the driver's psychological state may change further
from State 11+ to II. If the driver felt alarm, while observing the host vehicle came
close to the lead vehicle, his or her trust in the ACC system goes down a bit, and
a state transition occurs, say, from State II to II+ or from State II+ to III (see
Fig. 3.6) .

3.3.4 .2 Driver's Response to an Alert

Each psychological state is characterised by a corresponding driver performance
in situation recognition and swiftness of response to an emergency-braking alert .
A model for the driver 's response to an emergency-braking alert may be given as
follows : (I) If the driver was in State I when the alert was set off, he or she does
not respond to the alert at all. (2) If the driver was in State II, he or she stays in
the same state with probability p and hits the brake pedal in T2 seconds . With
probability 1 - p, the driver state jumps into State IV. (3) If the driver was either
54 Inagaki

in State II+ or III, he or she applies the emergency brake himself or herself either
in T2+ or T3 seconds, respectively. (4) In State IV, the driver panics and fails to
take any meaningful actions to attain car safety. T2, T2+ and T3 are treated as
random variables with different means.

3.3.4.3 Evaluation of Efficacy of Levels of Automation

The following are realistically feasible design alternatives for cases in which the
target vehicle makes a rapid deceleration at a rate much greater than the maximum
deceleration rate which the ACC system can handle with its ordinary automatic
brake:
Scheme 1: Upon recognition of a rapid deceleration of the target vehicle, the ACC
system gives an emergency-braking alert, where the LOA is positioned at 4.
Scheme 2: Upon recognition of a rapid deceleration of the target vehicle, the ACC
system gives an emergency-braking alert. If the driver does not respond within
a pre-specified time (due to inattentiveness or delay in situational recognition),
it applies an automatic emergency brake, where the LOA is positioned at 6.
Scheme 3: Upon recognition of a rapid deceleration of the target vehicle, the ACC
system applies its automatic emergency brake simultaneously when it issues an
emergency-braking alert, where the LOA of this scheme is positioned at 6.5.
With the models of driver's psychological state and response time for each state,
Monte Carlo simulations can be performed to evaluate efficacy of the LOAs under
possibility of driver's overtrust in the ACC system. The following are some of
observations obtained (Inagaki and Furukawa, 2004):
(1) A safety control scheme with LOA-6 may not be effective, compared to one
with LOA-4, if the driver trusts in the automation excessively. In order to
mitigate the drawback of the LOA-6, some measures are needed to keep the
driver alert.
(2) The drawback of LOA-6 is partially due to the time delay of the automatic
safety control action. The LOA-6.5 scheme may be more effective than LOA-6
or LOA-4 in order to assure systems safety in time - criticality under possibility
of driver's complacency.

3.4 Concluding Remarks

This paper has given brief descriptions on two driver behaviour modelling re-
lated research projects in Japan the aims of which are to develop proactive safety
technologies to detect mismatches between driver's behaviours and the traffic en-
vironment. This paper discussed two models in the projects: one is the Bayesian
network model for detecting non-normative behaviour of the driver and the other
is the discrete-event model for analysing driver interactions with 'smart and reli-
able' automation. These models may still be in their early stages of development,
3. Towards Monitoring and Modelling for Situation-Adaptive Driver Assist Systems 55

compared to control theoretic models such as those discussed by Weir and Chao
(2005) and Juergensohn (2005). However, the models are expanding their descrip-
tion capabilities and applicability.
Drivers must be provided with necessary and sufficient supports by machine
or automation. Intention understanding and communication play important roles
in realising such meaningful support functions. If the driver fails to understand
the intention of machine intelligence, an automation surprise may happen. If the
machine does not recognise the driver's intention, its 'good support' may be an-
noying to the driver. The topic of intent inference attracts keen interests in Japan as
well as the rest of the world. Although no discussion could be made in this paper,
a fuzzy association system model with case-based reasoning has been developed
for recognising human intention through monitoring his or her behaviour in the
'Humatronics' project in Japan for the safety of drivers and pedestrians (Umeda
et al., 2005; Yamaguchi et al., 2004).
In spite of its usefulness, an intention understanding approach may have some
limitations, at least at the present time. The second project described in this pa-
per, for instance, sometimes found difficulty in understanding the intention of the
driver from his or her behaviour. Norman (1988) has distinguished seven stages
of action, in which the first-four stages are (1) forming the goal, (2) forming the
intention, (3) specifying an action and (4) executing the action. The project tries to
infer the intention of the driver by catching some 'precursor' to the action (recall,
Fig. 3.3) that may come between stages (3) and (4). Some of problems observed
in the project are as follows: (a) No useful precursor may exist for some intended
actions, which might also be individual-dependent. (b) The driver may form a goal
for a very immediate future in dynamically changing environment, which makes
it hard either to infer its associated intention or to identify in a timely manner a
driver support that is appropriate for the intention. (c) When the driver's action
was corrected by the machine, the driver is likely to perceive that the machine did
not understand his or her intention at all, although the action needed to be cor-
rected because its execution timing did not match the traffic environment. Proactive
safety technologies usually assume machine intelligence. Failure in mutual under-
standing of intentions between the driver and the machine intelligence may bring
various inconveniences, such as automation surprises, distrust and overtrust. The
second project is now trying to challenge these problems via a context-dependent
adjustment of LOA as well as designing human interface that enables the driver
to (a) understand the rationale why the automation thinks so, (2) recognise in-
tention of the automation, (3) share the situation recognition with the automa-
tion and (4) perceive limitations of automation's functional abilities (Inagaki,
2006).

Acknowledgments. This study was conducted as part of the government project,

'Situation and Intention Recognition for Risk Finding and Avoidance' with the
support of the Ministry of Education, Culture, Sports, Science and Technology,
56 Inagaki

Government of Japan. The authorexpresses his appreciation to Dr.Cacciabue and

anonymous refereesfor theirconstructive comments and suggestions. Theyhelped
a lot in improving the quality and clarity of the paper.

References
Akamatsu, M. (2003). Measurement technologies for driver and driving behavior. In Pro-
ceedings oflEA 2003 (CD-ROM).
Akamatsu, M. and Sakaguchi, Y. (2003). Personal fitting driver assistance system based on
driving behavior model. In Proceedings of the lEA 2003 (CD-ROM).
Akamatsu, M., Sakaguchi, Y. and Okuwa, M. (2003). Modeling of driving behavior when
approaching an intersection based on measured behavioral data on an actual road. In
Proceedings of the Human Factors and Ergonomics Society 47th Annual Meeting (CO-
RaM).
Amiditis, A., Lentziou, Z., Polychronopoulos, A., Bolovinou, A. and Bekiaris, E. (2005).
Real time traffic and environment monitoring for automotive applications. In L. Macchi,
C. Re and P.C.Cacciabue (Eds.). Proceedings ofthe International Workshop on Modelling
Driver Behaviour in Automotive Environments (pp. 125-131).
Cacciabue, P.C. and Hollnagel, E. (2005). Modelling driving performance: A review of
criteria, variables and parameters. In L. Macchi, C. Re and P.C. Cacciabue (Eds.). Pro-
ceedings of the International Workshop on Modelling Driver Behaviour in Automotive
Environments (pp. 185-196).
Fuller, R. and Santos, J. (2002). Psychology and the highway engineer. In R. Fuller
and J. Santos (Eds.). Human Factors for Highway Engineers (pp. 1-10). Pergamon,
Amsterdam.
Furugori, S., Yoshizawa, N., Iname, C. and Miura, Y. (2003). Measurement of driver's
fatigue based on driver's postural change. In Proceedings ofthe SICE Annual Conference
(pp. 1138-1143).
Furugori, S., Yoshizawa, N., Iname, C. and Miura, Y. (2005). Estimation of driver fatigue
by pressure distribution on seat in long term driving. Review ofAutomotive Engineering,
26(1), 53-58.
Green, M. (2003). What causes 90% of all automobile accidents? Available at
http://www.visualexpert.com/accidentcause.html.
Hashimoto, K. (1984). Safe Human Engineering. Chuo Rodo Saigai Boushi Kyokai (in
Japanese).
Hollnagel, E. and Bye, A. (2000). Principles for modeling function allocation. International
Journal ofHuman-Computer Studies, 52, 253-265.
Inagaki, T. (2003). Adaptive automation: Sharing and trading of control. In E. Hollnagel
(Ed.). Handbook of Cognitive Task Design (pp. 147-169). Lawrence Erlbaum, Mahwah,
NJ.
Inagaki, T. (2006). Design of human-machine interactions in light of domain-dependence
of human-centered automation. Cognition, Technology and Work, 8,161-167.
Inagaki, T. and Furukawa, H. (2004). Computer simulation for the design of authority in
the adaptive cruise control systems under possibility of driver's over-trust in automation.
In Proceedings of the IEEE SMC Conference (pp. 3932-3937).
Inagaki, T., Moray, N. and Itoh, M. (1998). Trust self-confidence and authority in human-
machine systems. In Proceedings of the IFAC Man-Machine Systems (pp. 431-436).
3. Towards Monitoring and Modelling for Situation-Adaptive Driver Assist Systems 57

Itoh, M., Akiyama, T. and Inagaki, T. (2006). Driver behavior modeling. Part II: Detec-
tion of driver's inattentiveness under distracting conditions. In Proceedings of the DSC-
Asia/Pacific (CD-ROM).
Juergensohn, T. (2005). Control theory models of the driver. In L. Macchi, C. Re and
P.C. Cacciabue (Eds.). Proceedings of the International Workshop on Modelling Driver
Behaviour in Automotive Environments (pp. 37-42).
Miura, Y., Yoshizawa, N. and Furugori, S. (2002). Individual variation analysis of body
pressure distribution for long term driving simulation task. The Japanese Journal of
Ergonomics, 38(Suppl), 70-72.
Norman, D.A. (1988). The Psychology of Everyday Things. Basic Books, New York.
Panou, M., Bekiaris, E. and Papakostopoulos, V.(2005). Modeling driver behavior in EU and
international projects. In L. Macchi, C. Re and P.C. Cacciabue (Eds.). In Proceedings of
the International Workshop on Modelling Driver Behaviour in Automotive Environments
(pp.5-21).
Saad, F. (2005). Studying behavioural adaptations to new driver support systems. In L.
Macchi, C. Re and P.C. Cacciabue (Eds.). Proceedings of the International Workshop on
Modelling Driver Behaviour in Automotive Environments (pp. 63-73).
Sakakibara, K. and Taguchi, T. (2003). Biochemical measurement for driver's state based
on salivary components. In Proceedings of the lEA 2003 (CD-ROM).
Sakakibara, K. and Taguchi, T. (2005). Evaluation of driver's state of tension. In Proceedings
of the HCI International (CD-ROM).
Scerbo, M.W. (1996). Theoretical perspectives on adaptive automation. In R. Parasuraman
and M. Mouloua (Eds.). Automation and Human Performance (pp. 37-63). Lawrence
Erlbaum, Mahwah, NJ.
Sheridan, T.B. (1992). Telerobotics, Automation, and Human Supervisory Control. MIT
Press, Cambridge, MA.
Shiomi, K. and Hirose, S. (2000). Fatigue and drowsiness predictor for pilots and air traffic
controllers. In Proceedings of the 45th Annual ACTA Conference (pp. 1-4).
Taguchi, T. and Sakakibara, K. (2005). Evaluation of driver's state of tension. Review of
Automotive Engineering, 26(2), 201-206.
Tango, F. and Montanari, R. (2005). Modeling traffic and real situations. In L. Macchi, C.
Re and P.C. Cacciabue (Eds.). Proceedings of the International Workshop on Modelling
Driver Behaviour in Automotive Environments (pp. 133-147).
Treat, J.R., Tumbas, N.S., McDonald, S.T., Shinar, D., Hume, R.D., Mayer, R.E., Stansifer,
R.L. and Castellan, N.J. (1979). Tri-level study of the causes of traffic accidents: Final
report volume 1. Technical report, Federal Highway Administration, US DOT.
Umeda, M., Yamaguchi, T. and Ohashi, K. (2005). Intention recognition system using
case-based reasoning. In Proceedings of the RISP International Workshop on Nonlinear
Circuit and Signal Processing (pp. 243-246).
Weir, D.H. and Chao, K.C. (2005). Review of control theory models for directional and speed
control. In L. Macchi, C. Re and P.C. Cacciabue (Eds.). Proceedings ofthe International
Workshop on Modelling Driver Behaviour in Automotive Environments (pp. 25-36).
Witt, G.J. (2003). SAfety VEhicle(s) Using Adaptive Interface Technology (SAVE-IT) Pro-
gram, DTRS57-02-R-20003. U.S. Department of Transportation.
Yamaguchi, T., Matsuda, S., Ohashi, K., Ayama, M. and Harashima, F. (2004). Humane
automotive system using driver and pedestrian intention recognition. In Proceedings of
the World Congress on Intelligent Transport Systems (CD-ROM).
 II
Conceptual Framework and
Modelling Architectures
4
A General Conceptual Framework for
Modelling Behavioural Effects of
Driver Support Functions
lORAN ENGSTROM AND ERIK HOLLNAGEL

4.1 Introduction
In recent years, the number of in-vehicle functions interacting with the driver
has increased rapidly. This includes both driving support functions (e.g. anti-lock
brakes, collision warning systems, adaptive cruise control) and functions support-
ing non-driving tasks, e.g. communication and entertainment functions. Today,
many of these functions are also featured on portable computing systems, com-
monly referred to as nomadic devices. Moreover, in order to handle this growth in
diversity and complexity of in-vehicle functionality, several types of meta func-
tions for human-machine interface integration and adaptation have been proposed.
Such functions, often referred to as workload managementfunctions, are intended
to resolve potential conflicts between individual functions with respect to their
interaction with the driver (see Engstrom et al., 2004; Brostrom et al., 2006).
The term driver support functions will henceforth be used to refer to in-vehicle
functions that support what drivers do, whether related to driving or not.
The proliferation of driver support functions naturally changes the nature of
driving and the different types of functions may induce a variety of behavioural
effects. One general reaction to new driver support functions is that drivers change
their behaviour in various ways to incorporate the functions into the driving task,
an effect commonly referred to as behavioural adaptation (Smiley, 2000; Saad
et al., 2004). Another issue that has received much recent attention is the effects of
multitasking while driving, e.g. driver distraction due to interaction with in-vehicle
functions, passengers or other objects in the vehicle.
While a wide range of driver models exists, addressing different aspects of driver
behaviour, a generally agreed conceptual framework for describing behavioural ef-
fects of driver support functions is still lacking. Technological and methodological
development in this area is therefore generally made without a common concep-
tual basis. The objective of this paper is to outline some basic requirements for
such a conceptual framework as well as to propose a specific candidate. The main
starting point will be models based on the cognitive systems engineering tradition,
specifically the COCOMIECOM framework (Hollnagel et al., 2003; Hollnagel and
Woods, 2005). It should be stressed that the aim here is not to present a validated

61
62 Engstrom and Hollnagel

model of driver behaviour but rather to propose a general conceptual framework

that can be used to describe behavioural effects of driver support functions, and
how these effects relate to accident risk. A principal motivation for this work was
the need for a common conceptual framework in the AIDE (Adaptive Integrated
Driver-vehicle Interface) ED-funded project, which deals with technical and hu-
man factor issues related to driver behaviour and automotive human-machine
interface development (see Engstrom et al. (2004) for an overview of the AIDE
project).
The chapter is organised as follows: In the next section, the general intended ap-
plication areas and the key requirements for the framework are outlined. Section 4.3
provides a review of existing driver behaviour models that are relevant for present
purposes. In Section 4.4, the key elements of the proposed framework are de-
scribed. In Section 4.5, some specific example applications are described. Finally,
Section 4.6 provides a general discussion and conclusions.

4.2 Intended Application Areas and Requirements

In order to derive the basic requirements for the intended conceptual framework, it
is necessary to first consider how it is intended to be used. As mentioned above, the
main purpose of the framework is not a validated model of driver behaviour. Rather,
the idea is that the framework should provide a common language to describe key
issues related to behavioural effects of in-vehicle functions. More specifically, the
framework should be applicable to (at least) the following problems:

4.2.1 Functional Characterisation of Driver

Support Functions
As in many other fields, the development of driver support functions is still to a
large extent driven by technological possibilities rather than the actual needs of
the users. This often results in 'solutions that are looking for a problem' (Holl-
nagel, 2006). As a result, driver support functions are often described in terms of
the underlying technology, e.g. 'driver monitoring' and 'vehicle-to-vehicle com-
munication', while it is not always entirely clear what driver goals or task they
are intended to support. In order to link driver support functions to their potential
behavioural effects, it is useful to be able to characterise the different functions
with respect to their intended purpose, i.e. which goal(s) they support (including
driving- as well as non-driving-related goals). The present framework should be
able to provide such a functional taxonomy.

4.2.2 Coherent Description ofExpected Behavioural Effects

of Driver Support Functions
The framework should be applicable as a coherent conceptual basis to describe all
different types of behavioural effects of driver support functions, from the effect of
 4. A General Conceptual Framework for Modelling Behavioural Effects 63

multitasking while driving to short- and long-term behavioural adaptations such

as risk compensation and over-reliance on the support function.

4.2.3 Conceptualising Relations Between Behavioural

Effects and Road Safety
While a basic understanding of behavioural effects of driver support functions is
an important goal in itself, a further critical issue concerns the relation between
such effects and actual road safety, e.g. changes in incident and accident risk
(see Dingus (1995) for a good discussion on this topic). While not intended as
a general accident model, the framework proposed here should be applicable to
these problems as well.

4.2.4 Specific Requirements

From these three general intended application areas, some more specific require-
ments for the framework can be outlined. First, since different functions support
different aspects of driving, from low-level vehicle handling to high-level naviga-
tion and route planning, the framework must be able to account for behaviour on
all levels and, equally important, the relation between the different levels. Second,
related to the previous requirement, it is important that the same principles of
analysis are applied on all levels of description, i.e. the model should be recursive
(Cacciabue and Hollnagel, 2005). Finally, the framework must be able to account
for time, i.e. the dynamics of behaviour. This is particularly important to capture
the self-paced, adaptive, nature of driving.

4.3 Existing Models of Driver Behaviour

Driver behaviour modelling has a long history and a wide range of existing mod-
els address different aspects of driving. However, models specifically targeting
behavioural effects of driver support functions are rare. According to Michon
(1985), a general distinction could be made between taxonomic and functional
models. The former refers to descriptive models, or 'inventory of facts', without
an account of the interaction between the model components, including trait mod-
els (for example of driver accident proneness) and task analysis. The latter refers to
models that account for processes and/or interactions within the modelled system.
The focus of this review is on functional models.

4.3.1 Manual Control Models

The early driver modelling efforts focused mainly on control-theoretic models of
vehicle handling (e.g. Weir and McRuer, 1968). In these models, vehicle control
was generally modelled in terms of feedback control mechanisms with the goal
of minimising the difference between a reference (or target) state (e.g. the desired
heading angle) and the actual state. Later developments of these models included
64 Engstrom and Hollnagel

elements of feed forward control (McRuer et aI., 1977; Donges, 1978). While these
types of models are useful for modelling manual lateral and longitudinal vehicle
control in constrained situations, they do not capture higher level aspects of driving
such as decision making, planning or motivation.

4.3.2 Information Processing Models

During the past 40 years, the information processing paradigm, based on the digital
computer as the main metaphor, has dominated human factors and cognitive sci-
ence. The basic idea behind this paradigm is that human cognition can be modelled
as sequences of logically separated computational steps, including perception, de-
cision and response selection. Human attention and performance limitations are
then modelled in terms of limited capacity at these different stages (Moray, 1967;
Kahneman, 1973). The information processing paradigm has had a great influence
on theories of multiple task sharing, which are often used to understand the ef-
fects of interacting with a driver support system while driving. In particular, the
concept of mental workload, defined by de Ward (1996) as 'the specification of
the amount of information processing capacity that is used for task performance'
(p. 15), is directly based on the limited capacity metaphor. Most experimental work
in this area also involved dual task studies to investigate the level of interference
between different types of tasks, leading to the influential multiple resource theory
(Wickens, 2002). However, with a few exceptions (e.g. Shinar, 1992; Salvucci,
2001) information processing models have not been incorporated in more general
driver behaviour models. Whether this should be considered as a failure of the
automotive human factors trade, as proposed by Michon (1985), or as due to the
fact that 'early information processing models and their associated experimen-
tal techniques were incompatible with the requirements of complex tasks such
as driving' (Ranney, 1994), could be discussed. In any case, one shortcoming of
information processing models when applied in the driving domain is the basic
notion of the human as primarily a passive receiver of information, which makes
it difficult to account for drivers' active management of traffic situations, e.g. by
means of self-pacing and dynamic task allocation.

4.3.3 Motivational Models

Many existing models characterise driver behaviour in terms of dynamic regula-
tion of risk. These types of models are often referred to as motivational models
(Ranney, 1994). By contrast to information processing models, motivational
models emphasise the self-paced nature of driving and attempt to understand
the dynamical adaptation to varying driving conditions. The main differences
among existing motivational models concern the criteria proposed to govern the
adaptation. For example, the risk-homeostasis theory (Wilde, 1982) hypothesises
that drivers strive to maintain a constant level of accepted risk. By contrast, the
zero-risk theory (Summala, 1985; 1988) states that the driver aims to keep the
subjectively perceived risk at zero-level. A related account has been offered by
Fuller (1984), who suggests that drivers' behaviour is guided by threat avoidance.
 4. A General Conceptual Framework for Modelling Behavioural Effects 65

In more recent model, the same author proposes that the driver attempts to maintain
a certain level of task difficulty rather than a level of risk (Fuller, 2005; see also
Chapter 10 in this book). Finally, based on Damasio's concept of somatic markers
(Damasio, 1994), Vaa (Chapter 12 in this book) proposes that driver behaviour is
largely driven by emotional responses to risky situations.
Motivational models have been criticised for being too unspecific regarding
internal mechanisms and, as a result, being unable to generate testable hypotheses
(Michon, 1985). It could also be argued that the need to include motivational or
emotional aspects partly is an artefact of the limitations of information processing
models. Since 'cold' cognition automatically excluded 'hot' cognition (Abelson,
1963), something important was missing from these models. This is, however, not
an issue in models that pre-dates information processing models, such as Gibson
and Crooks (1938, see below).

4.3.4 Safety Margins

As described in the previous section, a key issue in understanding driver behaviour
concerns the criteria that drivers use as the reference for adaptation. While the
motivational models use qualitative criteria such as risk or task difficulty, other
models have attempted a more quantitative approach based on safety margins. A
key starting point here is that humans, in most situations, tend to act as satisfiers
(rather than optimisers), i.e. they generally do not put more effort into a task than
needed (Simon, 1955). Many driver models incorporate the concept of subjectively
chosen safety margins as key criteria for guiding driver behaviour. Probably the
first account of safety margins in driving was offered by Gibson and Crooks (1938),
who proposed that drivers aim to stay within a 'field of safe travel', which can be
conceptualised as 'tongues' stretching out in front of the vehicle, with their size
and form being determined by the time-to-contact to surrounding obstacles. More
recently, this concept has been developed into more concrete time-based safety
margins parameters. There is abundant evidence that time-to-object information is
used by humans and animals for guiding locomotion (Gibson, 1979). Lee (1976)
used the perceptual variable tau, representing time-to-contact in terms of optically
specified parameters, to model drivers' braking behaviour. In traffic research, time-
to-collision is often used as a driving performance metric (van der Horst, 1990; van
der Horst and Godthelp, 1989; Minderhoud and Bovy, 2001). The corresponding
metric for lateral control is time-to-line-crossing (TLC). Godthelp et al. (1984)
demonstrated that TLC correlates strongly with driver's self-chosen occlusion
time. Too small TLC values are thus strong indicators of violations of the driver's
subjectively chosen safety margins. For a general account of safety margins in
driving, see Nilsson (2001).
Summarising the models reviewed in the present and the previous section, there
seems to be a strong convergence towards the general idea that driver performance
could be understood in terms of adaptation governed by some type of safety mar-
gins, although these are conceptualised differently by different authors, e.g. in
terms of objective quantitative parameters such as TTC and TLC, or more con-
cepts such as perceived risk or task difficulty.
66 Engstrom and Hollnagel

4.3.5 Hierarchical Models

A common approach is to model the driving task as a set of hierarchically organised
sub-tasks. A general hierarchical account of human performance is Rasmussen's
three-level model, which proposes a distinction between knowledge-based, rule-
based and skill-based performance (Rasmussen, 1983). Knowledge-based perfor-
mance is mainly needed in situations that have not been encountered before and
thus requires conscious deliberation, while rule-based performance refers to the
application of learned rules. Finally, skill-based performance refers to automated
skills that do not require any cognitive processing.
There are also several examples of hierarchical models that describe the driv-
ing task, e.g. Alexander and Lunenfeld (1986) who characterise the driving tasks
in terms of three levels: navigation, guidance and control. Another example is
Michon's influential description of the driving task in terms of strategic, tactical and
operational levels (Michon, 1985). These models, which have been called 'second
generation motivational models' (Ranney, 1994), describe goals and motives on
different levels, from general trip planning (strategic), via obstacle avoidance (tac-
tical) to immediate vehicle control (operational). It is assumed that activities on the
different levels interact dynamically but none of these models say much about how
this is actually accomplished. Hale et al. (1990) proposed to combine Michon's and
Rasmussen's levels into a two-dimensional matrix where (Rasmussen's) driving
performance levels are mapped onto (Michon's) driving task levels. Such a repre-
sentation is useful to describe differences between experts and novices in terms of
the degree to which different tasks in the hierarchy are automated (Ranney, 1994).
In most of these hierarchical driving models the different levels are (more or less
explicitly) intended to represent levels of information processing. A different type
of hierarchical driver model is the Extended Control Model (ECOM) proposed by
Hollnagel et al. (2003), which puts a hierarchical description of driving-related
goals (similar to Michon's model) into a control theoretic framework. ECOM pro-
vides a representation of the performance (rather than information processing)
on different levels of the driving task and also offers an account of the dynamic
interactions between concurrent activities on the different levels. In the current
version, four layers are proposed: targeting, monitoring, regulating and tracking.
The model is an extension of the Contextual Control Model (COCOM; Hollnagel,
1993; Hollnagel and Woods, 2005), which provides a general account of the dy-
namical coupling between perception (or, more generally, situation assessment),
decision and action. The COCOM and ECOM models are further described in the
following section.

4.4 A Conceptual Framework

In this section, the proposed conceptual framework is outlined. The section starts
with a short discussion on the key concept of behaviour and a description of how
concepts from control theory can be used to model the dynamics of behaviour.
 4. A General Conceptual Framework for Modelling Behavioural Effects 67

The key elements of the proposed framework, the COCOM and ECOM models,
are then presented.

4.4.1 Driver Behaviour as Goal-Directed Activity

Interaction between the driver and vehicle functions can be viewed as an instance
of the more general notion of driver behaviour. Thus, it is important to define
more precisely what we mean by behaviour. As suggested, e.g. by Dennett (1987),
behaviour, as opposed to mere bodily movement, can be defined as goal-directed
activity. According to Dennett, we understand the activity of other people by
adopting what he calls the intentional stance. This entails ascribing intentions and
goals to other agents, which enables us to make predictions about their behaviour
(naturally, these ascriptions may not always correspond to actual intentions of
people).
Based on this notion, driver behaviour can be understood in terms of what is
required to accomplish a number of different goals while driving. In addition to
driving related goals (such as keeping within the lane or reaching a destination),
drivers may also be occupied with other goals that are only vaguely related, or
entirely unrelated, to vehicle operation (such as finding a track on an MP3 player).
Thus, driver behaviour could be defined as the general pursuit of driving- and
non-driving-related goals while driving. Consequently, the general role of driver
support functions is to support the driver in accomplishing these goals. Driving
can further be defined as the subset of driver behaviours aimed towards goals
associated with vehicle operation. The role of driving support functions (a subset
of driver support functions) is thus to support driving-related goals.
A distinction can also be made between goals that are permanent (such as avoid-
ing accidents, complying with the traffic code and avoiding risk) hence applicable
to more than one journey, persistent (i.e., valid for the duration of a single journey,
such as reaching the destination) and transient in the sense that they may come and
go during a journey (e.g., overtaking the car in front, reaching a cafeteria before
noon).

4.4.2 Dynamical Representation ofDriver Behaviour

In order to understand driver behaviour in terms of goal-directed activity, we need
a way to describe how drivers' goals are dynamically achieved and maintained.
Control is a useful concept for describing the dynamics of goal-directed behaviour,
in man as well as in machines. In general, control can be understood as the ability
to direct and manage the development of events (Hollnagel and Woods, 2005).
Controlling a process means that actions are determined by the aim to achieve a
consistent goal state (often called the reference or target value), e.g. by means of
countering effects of external disturbances. Control is thus closely related to order-
liness or predictability, i.e. a controlled system is orderly, stable and predictable
while a system that is out-of-control is disorderly, unstable and unpredictable.
68 Engstrom and Hollnagel

There are two basic forms of control: In feedback (or compensatory) control,
the controller performs corrective actions based on the deviation between a desired
outcome (the goal) and the actual state. The prototypical example of a feedback
control system is the thermostat. Another type of control is feed forward (or an-
ticipatory) control. In this case, control actions are based on predictions of future
states and, hence, proactive rather than reactive. Driving behaviour is generally a
mixture of feedback and feed forward control.
In engineering control applications, such as a thermostat or an Adaptive Cruise
Control system, the target values (i.e. the desired temperature and time gap re-
spectively) are generally determined beforehand by the engineer or set by the user.
However, when analysing human controlled behaviour, identifying the target, or
controlled variable, is often viewed as the key issue (Powers, 1998; Marken, 1986).
The indentification of the controlled variable is complicated by the fact that the
human controller, as mentioned above, seldom operates as an optimiser but rather
acts as a satisfier (Simon, 1955), thus tolerating a certain deviation from the target
value. In driving, time based safety margins (e.g. van der Horst and Godthelp,
1989) or risk thresholds (Summala, 1988) could be thought of as reference values
for the vehicle control loop (although this idea has not yet, to our knowledge, been
exploited in extisting models).
As described in the previous section, control theory has been widely applied
to the modelling of vehicle handling, in automotive and other domains (e.g. Weir
and McRuer, 1968; Donges, 1978). Control theory has also been applied to the
modelling of higher-level aspects of driving. For instance, many of the moti-
vational models reviewed above, especially the risk homeostasis model (Wilde,
1982), are based on control theoretical concepts. However, few existing models
allow for a unified representation of controlled behaviour at different levels of
the driving task. One existing modelling framework able to provide such descrip-
tions is the COCOMIECOM model (Hollnagel and Woods, 2005), which thus
has been selected as the main starting point for the proposed conceptual frame-
work. The next two sections describe the COCOM and ECOM models in more
detail.

4.4.3 The Contextual Control Model (COCOM)

The contextual control model (COCOM), described in Hollnagel (1993) and Holl-
nagel and Woods (2005), provides a general account for modelling human control
of a process or plant, based on Neisscr's (1976) perceptual cycle concept. An im-
portant starting point for COCOM is that the controller and controlled system is
viewed as a joint cognitive system (JCS). The central object of study is then the
JCS rather than the controller or the controlled system in isolation. This approach
offers a perspective that differs from traditional information processing models
where the human operator and the machine are normally treated as logically sep-
arate entities. The boundaries between the JCS and its environment are generally
determined by the objective of the analysis. Pragmatically speaking, an object is
 4. A General Conceptual Framework for Modelling Behavioural Effects 69

considered part of the JCS if (I) it is considered important for the ability of the
JCS to maintain control and (2) it can be controlled by the JCS. Moreover, objects
satisfying (2) but not (I) may be included in the JCS if this facilitates the purpose
of the analysis (Hollnagel and Woods, 2005). For present purposes, the JCS of
interest is in most cases the joint driver-vehicle system (JDVS).
COCOM is a general model of how control is maintained by a JCS and is appli-
cable across a range of different JCS types on different levels of description. This
type of analysis makes minimal assumptions about internal cognitive processes
and focuses on behaviour and the dynamical interactions between the compo-
nents in the JCS, in particular on how the JCS maintains, or loses, control of a
situation .
A central concept in COCOM is the construct-action-event cycle. The con-
struct refers to what the controller knows or assumes about the situation in which
the action takes place. The construct is the basis for selecting actions and in-
terpreting information. The selected actions affect the process/application to be
controlled. This generates events that provide feedback on the effects of the
action which, together with external disturbances, modifies the construct and,
hence, the future action selection . An important property of this model is thus
that it accounts for both the feedback and feedforward aspects of control, i.e. ac-
tion selection is a function of both direct feedback and the predictions of future
events.
The main factors determining the level of control maintained by a JCS is pre-
dictability and available time. A key property of COCOM is that it offers an
explicit account of time. Figure 4.1 gives an illustration of the three main temporal
parameters involved: (I) Time to evaluate events (T E) , (2) time to select an action
or response (T s) and (3) time to perform an action (T p ; see Hollnagel and Woods,

External event I
... ~ disturbance

Events l ...........
feedback
TE =time to
evaluate
events
Tp =time to
perform an
action

FiGURE 4.1. The contextual control model (Hollnagel and Woods, 2005) .
70 Engstrom and Hollnagel

2005, for a more elaborate description of the timing relations in COCOM). As

in any control system, the relations between these time parameters determine the
performance of the JCS. The relation between the time needed to evaluate, select
and perform an action (TE , Ts and Tp ) and the time available (TA ) is of special
importance. In order to maintain control, the time needed to perform one cycle
(TE + Ts + Tp ) must, in the long run, be less than the total time available. If the
available time becomes too small, it may be increased by slowing down the pace
of the task, e.g. by reducing speed in the case of driving (as long as the task is
self-paced). It is in many cases uncertain how much time is available, due to the
dynamics of the environment, including the unpredictability of other drivers. In
such cases humans tend to sacrifice thoroughness in order to maintain efficiency,
as described by the efficiency-thoroughness trade-off (ETTO) strategy (Hollnagel,
2004).

4.4.4 The Extended Control Model (ECOM)

While COCOM only describes a single control process (i.e. pursuit of a single
goal), the driving task generally involves the pursuit of several simultaneous sub-
goals with different time frames. A long-term goal could be to reach a destination
in time. An example of a medium-term scale goal is to overtake a vehicle ahead,
while short-term goals include staying in lane and avoiding obstacles. Note that
the temporal characteristics of goals is different from whether they are permanent,
persistent, or transient. A goal may also subsume goals on shorter time frames. For
example, in order to reach a destination in time, it may be necessary to overtake
a number of vehicles. This, in turn, requires safe vehicle handling in order to
avoid collisions. Thus, the driving task can be described as a set of simultaneous,
interrelated and layered control processes. In addition, drivers generally also pursue
goals that are unrelated to driving, e.g. talking to a passenger, using the cell phone,
looking for a place to eat, etc.
As reviewed above, this hierarchical organisation is reflected in many models of
driving, e.g. Alexander and Lunenfeld (1986) and Michon (1985). However, none
of these models provide a sufficient account of the dynamical aspects of driving
(i.e. the relation between performance and time), or the simultaneous relations
between control processes on different layers. This is offered by the Extended
Control Model (ECOM) (Hollnagel et al., 2003; Hollnagel and Woods, 2005),
which represents a multi-layered extension to COCOM. The basic structure of
ECOM is illustrated in Fig. 4.2.
A key assumption behind the ECOM model is that goals on different layers are
pursued simultaneously and that these goals and their associated control processes
interact in a non-trivial way. In the current version, four control layers are pro-
posed: Tracking, regulating, monitoring and targeting, where each layer potentially
contains multiple parallel control processes (it should be noted that these partic-
ular layers are not fixed and may be revised and adapted to different application
domains).
 4. A GeneralConceptual Framework for Modelling Behavioural Effects 71

Targeting ----

Anti cipatory
control

Tracking

FIGUR E 4.2. The extended control model (ECOM).

A key property of the model is that during normal (controlled) task performance,
the goals/targets for the control processes on a given layer are determined by the
control processes one layer up. In the driving domain, the tracking control refers to
the momentary, automated, corrections to disturbances, e.g. wind gusts . Regulat-
ing refers to more conscious processes of keeping desired safety margins to other
traffic elements. This determines the target values for the tracking control loops .
Monitoring refers to the control of the state of the joint vehicle-driver system rela-
tive to the driving environment. It involves monitoring the location and condition
of the vehicle, as well as different properties of the traffic environment, e.g. speed
limits. This generates the situation assessment that determines the reference for
the regulating layer. Finally, the targeting control level sets the general goals of
the driving task , which determines the objectives for the monitoring layer. The
functional characteristics of the four layers , in the context of driving, are described
in Table 4.1.
As illustrated in Fig. 4.2, tracking control is typically based on feedback (com-
pensatory control) while monitoring and targeting are mainly of the feed-forward
(anticipatory) type . The regulating layer may involve a mix of feedback and feed
forward control. The ECOM model provides an account of how goals at different
layers interact and how higher goals propagate all the way down to moment-to-
moment vehicle handling. Control tasks on different layers may also interfere with
each other and disturbances on lower layers may propagate upwards. For exam-
ple, looking for directions (monitoring) may disrupt visual feedback, which may
affect regulating and tracking control ; a sudden experience of slipperiness due to
ice on the road , manifested on the tracking level may modify higher-level goals
(e.g. increasing safety margins on regulating level) and even change the targets for
driving (e.g . choosing a different route) .
72 Engstrom and Hollnagel

TABLE 4.1. Functional characteristics of the EeOM layers.

Typeof control Demandsto Frequencyof Typicalduration
involved attention occurrence
Targeting Goal setting High, Low (mostly Short (minutes)
(feedforward) concentrated pre-journey)
Monitoring Condition Low (car). High Intermittentbut 10 min to duration
monitoring (traffic, regular (car). of voyage
(feedback + hazards) Continuous
feedforward) (traffic,
hazards)
Regulating Anticipatory High (uncommon Very high (town), 1 s - 1 min
(feedback + manoeuvres), medium
feedforward) Low (common (country)
manoeuvres)
Tracking Compensatory None Continuous <Is
(feedback) (pre-attentive)

4.5 Application
This section gives some examples of how the proposed framework can be applied
in the three problem domains stated in Section 2: (1) Characterisation of driver
support functions, (2) description of their behavioural effects and (3) reasoning
about the relation between behaviour/performance and road safety.

4.5.1 Characterising Driver Support Functions

In terms of the proposed framework, driver support functions are characterised
with respect to the goal(s) they are intended to support (where some functions
support non driving-related goals). Based on this, a tentative taxonomy of in-
vehicle functions could be outlined. The sub-set of driving support functions (with
'driving' defined as in Section 4.1) can be categorised with respect to the ECOM
layers described in the previous section. In addition, driver support functions
include non-driving related functions as well as workload management 'meta'
functions with the purpose to manage the different driving and non-driving related
functions with respect to their interaction with the driver. These function categories
are further described below, and illustrated in Fig. 4.3.

4.5.1.1 Support for Tracking

This category includes functions which support driving by (partly) automating
tracking control actions of speed and direction, e.g. antilock brakes (ABS), dy-
namic stability and traction control (DSTC), adaptive cruise control (ACC) and
lane keeping aid (LKA). For some functions, such as ACC, this means that human
involvement is to a large extent re-allocated to the regulating control layer. In fully
 4. A General Conceptual Framework for ModellingBehavioural Effects 73

Driver support functions

Driving support Non-driving

related functions

Targeting

Monitoring Speed alert

Regulating Co llision warning

Tracking ACC ABS

Worklo ad
management

FiG URE 4.3. Examplesofdriversupportfunctions mappedontotheECOMlayers. Workload

management can be viewedas a metafunctionthatarbitratesbetweendifferentcontroltasks.
(See the text for further explanation.)

automated driving, the driver's task would be limited to targeting, e.g. setting the
destination, perhaps supported by monitoring of non-instrumented information
sources.

4.5 .1.2 Support for Regulating

These are functions with the purpose to support regulatory control, for instance
helping the driver maintain adequate safety margins. This can be done , e.g. by
providing warnings when safety margins are about to be violated or otherwise
enhancing the perception of safety margins. Examples include forward collision
warning (FeW), lane departure warning (LDW) and night vision systems.

4.5.1 .3 Support for Monitoring

In-vehicle functions belonging to this category support the monitoring of higher
level aspects related to the driver-vehicle-environment state . Examples include
vehicle state monitoring functions (e.g. fuel and oil level indication), route guid-
ance and traffic information functions. Importantly, activities on the monitoring
level may interfere with tracking and regulatory control, as further discussed
below.

4.5 .1.4 Support for Targeting

Targeting is in driving, as well as in most other domains, predominantly a human

activity. To some extent a smart navigation system can be seen as impinging on
that , but on the whole, targeting goals are set by humans and not by machines.
74 Engstrom and Hollnagel

4.5.1.5 Non-Driving-Related Functions

In addition to the different types of driving support functions mentioned above,
there are many in-vehicle functions that primarily support other tasks than driving.
Examples include the radio, the media players, the phone and various functions
supporting the work task of professional drivers, e.g. fleet management functions.
However, it should be noted that such functions may be used to support driving,
e.g. in a situation where the mobile phone is used to obtain directions or a traffic
broadcast forces the driver to revise his/her goals. These functions are nevertheless
conceptually distinct from driving support functions by the fact that they are not
primarily intended to support driving.

4.5.1.6 Workload Management Functions

Workload management functions can be viewed as 'meta-functions', responsible
for coordinating individual functions, e.g. by means of information prioritisation
or by putting non-critical information on hold in demanding driving situations. So
far only a few systems of this type have entered the market (e.g. Volvo cars' intel-
ligent driver information system (IDIS) and Saab's dialogue manager), but more
advanced functions are being developed in different research efforts, both in-house
at the companies (see Brostrom et aI., 2006, for a description of the next generation
IDIS) and in collaborative efforts such as COMUNICAR (Amditis et aI., 2002),
AIDE (Engstrom et aI., 2004) and SAVE-IT (SAVE-IT, 2001). By contrast to the
individual functions described above, the workload management meta-functions
do not directly support specific driver goals. Rather, in terms of the proposed
framework, these functions could be conceptualised as indirectly supporting the
driver by resolving conflicts between different (driving- or non-driving related)
goals. One key objective is to promote safe driving by means of supporting the
driver in prioritising the tracking and regulating control tasks in demanding driving
situations. It should be noted that the adaptivity provided by workload manage-
ment functions, e.g. changing presentation format to resolve conflicts with other
functions or the driving situation, potentially reduces the predictability of system
behaviour. Since, as mentioned above, predictability is of key importance for the
JCS to maintain control, great care must be taken in the design of such functions
in order to avoid unexpected usability and/or safety problems.

4.5.2 Characterising Behavioural Effects of

Driver Support Functions
The behavioural effects of a specific driver support function is the result of a
complex dynamic interaction between individual driver characteristics (motivation
for driving, subjectively chosen safety margins, driving skills, personality, effort
etc.), vehicle parameters (e.g. steering and braking dynamics) and the driving
environment (road type, curvature, lane width, traffic density etc.). It is useful to
make a general distinction between direct and indirect (side) behavioural effects.
 4. A General Conceptual Framework for Modelling Behavioural Effects 75

Direct effects are those that are intended by the system designers and implied by the
system's functional specification. For driving support functions, direct effects are
normally the intended performance enhancements on one or more control layers
(e.g. increased lane keeping performance for LDW and increased route-finding
ability for navigation support). By contrast, indirect effects are not intended by
the designers (and thus not implied by the functional specification). It should be
noted that indirect effects are not necessarily bad. For example, a potential positive
indirect behavioural effect of a lane departure warning system could be an increased
use of the turn signal.
In the following sections, some of the main types of behavioural effects
found in the literature are discussed from the perspective of the proposed frame-
work.

4.5.2.1 Behavioural Adaptation to Driving Support Functions

The ability to adapt to changing circumstances is central to human (and animal)
behaviour. The term behavioural adaptation (BA) generally refers to 'the whole
set of behaviour changes that are designed to ensure a balance in relations between
the (human) organism and his surroundings and at the same time the mechanisms
and processes that underlie this phenomenon' (Bloch et al., 1999). However, in
traffic research, behavioural adaptation is often used to refer to a more specific
type of adaptation, as defined by OECD (1990): 'Those behaviours which may
occur following the introduction of changes to the road-vehicle-user system and
which were not intended by the initiators of the change.' (p. 23). Hence, accord-
ing to this definition, BA only refers to indirect effects, e.g. compensatory be-
haviours that may reduce, or even cancel out, the expected benefits of a safety
measure.
BA has been demonstrated for many different types of driver support functions
(see Smiley, 2000 and Saad et al, 2004, for examples). According to the proposed
framework, such changes can be more precisely described in terms of performance
on the ECOM control layers and the relation between them. For example, in one
of the most cited studies on BA, it was found that taxi drivers with ABS- (anti-lock
braking system) equipped-vehicles tended to drive with higher speed and adopt
shorter headways than drivers of vehicles without ABS (Fosser et aI., 1997). (This
particular effect was actually predicted almost 60 years earlier by Gibson and
Crooks (1938)). According to the proposed framework, this phenomenon can be
described in terms of a change of goals and criteria on the regulating layer, caused
by changing conditions on the tracking layer. When ABS is activated, drivers expe-
rience that the improved braking (i.e. longitudinal tracking) performance increases
their safety margins due to the shorter stopping distance. Thus, if there is a high-
level motivation (on the targeting level) for arriving quickly at the destination (as is
often the case for taxi drivers), the drivers can reduce headway and increase speed
while still keeping within their subjective safety margin. (Whether the actual risk
remains constant is, however, another issue). This safety margin threshold could
thus be viewed as the reference value for the tracking loop. This example shows
76 Engstrom and Hollnagel

how the dynamical interaction between high- and low level driving goals can be
conceptualised by the present framework.

4.5.2.2 Effects of Multitasking While Driving

An important issue that has attracted much recent attention is how the time sharing
between driving and other tasks affects driving performance and safety. A typical
case of such multitasking is the use of different in-vehicle information functions
while driving. These tasks are often referred to as secondary tasks while driving
is considered the primary task. The effects of secondary tasks are often concep-
tualised in terms of distraction, which has been defined as 'attention given to a
non-driving related activity, typically to the detriment of driving performance' (ISO
TC22/SCI3/WG8 CD 16673). However, this type of terminology creates concep-
tual difficulties because 'driving-related activity' and 'driving performance' are
not further defined. For instance, reading a map on the navigation system display
should clearly be regarded as a 'driving related activity'. On the other hand, it
could at the same time distract the driver and cause degraded driving performance.
Related problems have also been noted in analysis of naturalistic driving data in
the recent 100 car study. As pointed out by Neale et al. (2005):
'Historically, driver distraction has been typically discussed as a secondary
task engagement. Fatigue, has also been described as relating to driver inatten-
tion. In this study, it became clear that the definition of driver distraction needed
to be expanded to a more encompassing 'driver inattention' construct that in-
cludes secondary task engagement and fatigue as well as two new categories,
'Driving-related inattention to the forward roadway' and 'non-specific eye glance'.
'Driving-related inattention to the forward roadway' involves the driver checking
rear-view mirrors or their blind spots. This new category was added after viewing
multiple crashes, near crashes and incidents for which the driver was clearly pay-
ing attention to the driving task, but was not paying attention to the critical aspect
of the driving task (i.e. forward roadway) at an inopportune moment involving a
precipitating factor.' (p. 6).
The proposed framework reconciles these conceptual difficulties in a quite
straightforward way. As suggested above, the driving task should not be seen
as a single activity, but rather as a set of multiple simultaneous and layered control
tasks. Thus, driving performance could be defined with respect to any of these
control tasks. For example, driving performance on the tracking level is associated
with the ability to keep the vehicle within acceptable safety margins. Similarly,
performance on the regulating level would be related to the ability to select appro-
priate safety margins based on a general situation assessment at the monitoring
level. This situation assessment may induce inattention to the forward roadway
(e.g. when checking mirrors), and could hence be described as a distraction on the
tracking and/or regulating layers. In general, distraction with respect to a given
control process (e.g. tracking) could thus be viewed in terms of interference by
another (driving- or non-driving related) control process, typically resulting in
degraded performance on the given control task. Historically, 'driver distraction'
 4. A General Conceptual Framework for Modelling Behavioural Effects 77

often implicitly refers to interference with the tracking and/or regulating control
tasks (and this is probably the intended meaning of the definition cited above).
However, the present framework enables a more precise conceptualisation and
makes it possible to describe in more detail which tasks/control processes that are
affected in a given distraction scenario.
Another commonly used concept is mental workload. While distraction is de-
fined on the basis of attention allocation, driver mental workload refers to the
amount of resources that the driver needs to perform one or more tasks, relative
to a limited subjectively defined resource pool. As mentioned in section 4.3.2,
the concept has strong roots in the information processing paradigm (e.g. Moray,
1967). It should be noted that mental workload is not a necessary precondition
for distraction, since inappropriate attention allocation may be caused by low-
workload tasks as well, e.g. daydreaming, checking mirrors and looking at road
signs. One limitation of the traditional workload concept, based on the limited
capacity metaphor, is that it does not account well for the dynamics of self-paced
driving. Based on the COCOMIECOM model, a more dynamic view of workload,
viewing the driver as an active agent, can be outlined where the spare resources
for a control process can be viewed in terms of the difference between total time
available and the total time needed to perform the control loop (see Fig. 4.1). Thus,
for example, if the available time for the tracking control loop is reduced, e.g. due
to time sharing with another visually demanding task (e.g. entering a mobile phone
number) the driver can gain time by reducing speed. The mechanisms that drive
this type of adaptive behaviour can, again, be understood in terms of the balance
between higher-level goals (e.g. the desire to arrive in time, i.e. targeting) and
lower level goals (e.g. to keep acceptable safety margins, i.e. regulating).
In order to further illustrate the potential benefits of the proposed framework
compared to traditional information processing models, it is useful to take a closer
look at some empirical data from the HASTE EO-funded project. As part of
the project, a set of parallel experiments were conducted in different sites across
Europe, with the specific objective to investigate systematically the effects of visual
and cognitive load on driving performance and state (see, e.g. Engstrom et aI.,
2005a; Victor et aI., 2005; Jamson and Merat, 2005; Markkula and Engstrom, 2006;
Ostlund et al., 2004). In short, the HASTE results showed that visual time sharing
induced increased lane position variation, increased number of large steering wheel
reversals, reduced speed and increased headway to lead vehicles. By contrast, time
sharing with purely cognitive tasks (i.e. tasks that require no visual interaction)
did not interfere with tracking control at all, a result consistent with the meta-
analysis of mobile-phone studies made by Horrey and Wickens (2004). Rather,
results from HASTE indicate that cognitively loading tasks lead to significantly
improved tracking control in terms of reduced lane keeping variation compared to
baseline driving (e.g. Engstrom et aI., 2005a; Jamson and Merat, 2005), an effect
that has also been documented in other studies (e.g. Brookhuis et aI., 1991). This
increased lane keeping performance for cognitive tasks was also accompanied by
a concentration of gaze towards the road centre (Victor et aI., 2005), an effect also
found in previous studies (e.g. Harbluk and Noy, 2002; Recarte and Nunes, 2003)
78 Engstrom and Hollnagel

and an increased number of steering micro corrections (Markkula and Engstrom,

2006). Other studies have also found that cognitive load impairs signal detection
performance (e.g. Greenberg et al., 2003; Engstrom et al., 2005b). Finally, speed
adaptation is seldom observed for cognitive tasks (Ostlund et al., 2004) and some
of the HASTE experiments found that the longitudinal safety margin, in terms of
time-headway to a lead vehicle, was even reduced during cognitive load (Jamson
and Merat, 2005).
A limited capacity model, viewing workload as the single factor that modifies
behaviour (i.e. 'high workload leads to worse performance'), is of limited use
for understanding these behavioural patterns. For instance, it would be hard to
make sense of the fact that lane keeping variation is increased for visual tasks
but reduced for cognitive tasks. In terms of the present framework, the effects of
the visual task can be understood as a degradation of lateral tracking control due
to reduced visual input during gazes away from the road. However, this is com-
pensated for, at least to some extent, on the regulating layer, by means of speed
reduction. By contrast, cognitive load does not interfere with tracking, but rather
seems to reinforce the lateral tracking performance (for possible explanations for
this peculiar phenomenon, see Engstrom et al., 2005a and Victor, 2005). In terms
of ECOM, the main interference of cognitive tasks seems to be on the regulating
and monitoring layers, as evidenced by the reduced detection performance, the
inability to adapt speed and the reduced time headway. It should be stressed that
the COCOMIECOM framework, as applied here, does not offer a detailed expla-
nation of the mechanisms underlying these phenomena. Rather, its main role is to
provide a suitable common language for describing the effects. The development
of more detailed explanations and models of multitasking while driving, and other
behavioural effets, is an important area for further research.

4.5.3 Driver Behaviour and Accident Risk

Understanding the relation between behavioural effects and actual accident risk is
one of the most important, and difficult, issues in current traffic safety research.
The difficulties are due to a lack of sufficiently detailed behavioural data in existing
accident databases as well as the lack of appropriate behavioural models. A further
problem is that the usual interpretation of risks assumes both decomposability and
linearity. Yet both of these qualities are absent in dynamic, complex environments
such as traffic.
Based on the proposed framework, the relation between driving performance
and risk (for an individual driver) can be thought of as a complex function involving
(at least) the following factors:

1. The current complexity/difficulty of the driving task.

2. The driver's vehicle handling skills (performance on the tracking layer).
3. The ability to make a correct situation assessment (performance on the moni-
toring layer).
 4. A General Conceptual Framework for Modelling Behavioural Effects 79

4. The ability to adopt safety margins that are appropriate to (1) and (2), based on
(3) (determines performance on the regulation layer).
5. The effort spent on the control tasks on the different layers (2-4).

Based on this view, it is clear that performance degradation on one control layer
does not automatically increase accident risk. For example, reduced tracking per-
formance (e.g. due to visual distraction) is most risky if not properly compensated
for on the regulating layer (e.g. by reducing speed). Thus, risk (as a function of
performance) must be understood in terms of the relation between performance
on the different layers, where inadequate adaptation to the current driving condi-
tions and ones own driving skills could be hypothesised to be a critical factor. A
typical example of this is drunk driving, where alcohol is well known to induce
overestimation of the own performance driving capability. Thus, in this case, the
erroneous safety margin setting (on the regulation layer) is due to the cognitive
impairment induced by the drug. This line of reasoning also applies to individ-
ual differences with respect to risk taking. For example, the over-involvement of
young male drivers in accidents could be understood as in terms of overestimation
of driving performance combined with a higher propensity for sensation seeking
(i.e. different goals on the targeting level), leading to inadequate safety margin
settings. Yet another example is run-off-road accidents due to slippery roads. In
this case, the erroneous safety margin setting (reflected, e.g. in too high speed
in a curve) is due to an erroneous situation assessment on the monitoring layer,
which affects the regulatory level and finally induces instability and breakdown of
the tracking control. As discussed in the previous section, there is evidence that
cognitively loading tasks, such as phone conversation, impairs the ability to set ap-
propriate safety margins and adapt accordingly. However, the safety consequences
of this are still unknown.
It is very difficult to determine whether adequate adaptation has been achieved
in a particular situation, especially in terms of quantitative driving performance
metrics. One potential approach is to look for violations ofsafety margins. Possible
metrics of such violations include the amount of involuntary lane departures, min-
imum TLCITTC value, the amount of TLC/TTC values below some critical value
(van der Horst, 1990), or the total time spent below the critical value (time-exposed
TTC-TET; Minderhoud and Bovy, 2001). However, a basic problem with this
is that the accepted safety margins generally differ substantially between drivers
(and possibly also varies over time for an individual driver). Moreover, there is
yet no hard empirical data showing how these metrics relate to actual accidents.
Other key factors related to accident risk are expectancy and predictability (Victor,
2005). The ability to predict is central to remain in control and the occurrence of
unexpected events increases the risk for losing control. An issue of key impor-
tance is the ability of the driver to regain control. This is more difficult if the loss
of control also involves the loss of goals (e.g. on the regulating level).
While the COCOMIECOM framework is suitable for describing and reasoning
about these issues, it is clear that accident causation cannot be explained only in
terms of inadequate adaptation, or reduced predictability, but rather in terms of the
80 Engstrom and Hollnagel

interaction of multiple (behavioural, technological or organisational) precipitating

factors. This requires models that deal specifically with accident causation, e.g.
models based on Human Reliability Analysis (e.g. Hollnagel, 1993, 1998). How-
ever, the COCOMIECOM framework is a useful starting point for describing and
analysing the behavioural factors involved in accident causation.

4.6 Discussion and Conclusions

The objective of the present paper was to outline a conceptual framework for
describing behavioural effects of driver support functions. A key starting point
was the view of behaviour as a goal-directed, situated and dynamic activity. Driver
behaviour involves the simultaneous pursuit of multiple goals that may be more
or less related to driving itself. The COCOMIECOM hierarchical control model
(Hollnagel and Woods, 2005) was proposed as the basis for the framework.
It was shown that the framework offers a coherent taxonomy for categorising
a wide range of driver support functions with respect to their functional purpose
(rather than the underlying technology). This could be very useful, e.g. for clearly
defining the scopes of different development and evaluation methodologies and
standards (such as, e.g. the European Statement of Principles on Human Machine
Interface; Commission of the European Communities, 2000) in terms of the types
of functions that they apply to.
The second intended application area was the conceptualisation of behavioural
effects of driver support functions. It was demonstrated that the framework yields
coherent descriptions of a range of different effects, from behavioural adaptation
to the effects of multitasking. Moreover, it was argued that the COCOMIECOM
framework accounts for many aspects that are missed when a traditional informa-
tion processing/workload-based perspective is adopted, for example the radically
different effects of visual and cognitive secondary task load on driving perfor-
mance.
The third intended application area concerned the understanding of accident
causation. It was illustrated how the framework could provide a useful starting
point for addressing the difficult question of how driving behaviour/performance
is related to risk. A main conclusion was that accident risk cannot generally be
explained by a single aspect of driving performance. According to the proposed
framework, risk needs to be understood in terms of the relation between different
levels of performance. Thus, for example, a certain amount of weaving in the lane
may not be too risky as long as it is appropriately compensated for by slowing
down or increasing headway. It should also be noted that a lack of behavioural
change, e.g. a failure to reduce speed when entering a slippery road segment, may
be safety critical as well. Thus, inadequate adaptation can be proposed as one key
behavioural factor related to risk.
The present framework is based on the view of the human driver as an active
agent that, in most situations, is able to maintain control of the vehicle. A key
advantage of the framework is that it is well suited to describe adaptive behaviour,
which is a prevalent effect of all types of driver support functions. This contrasts
 4. A General Conceptual Framework for Modelling Behavioural Effects 81

to the traditional information processing models, which tend to view the human
as a passive receiver of information, subject to overload if the limited capacity is
exceeded. Another key property of the framework is the characterisation of driving
performance in terms of hierarchical control. The idea of hierarchical driver models
is certainly not new and the ECOM does not contradict existing models such as,
e.g. Alexander and Lunenfeld (1986) and Michon (1985). However, the added
value brought by ECOM is that it offers an account for how the different layers are
related, e.g. how a change in a high-level goal (such as realising that one is late to
the airport) can change performance on lower control layers (such as reducing the
accepted safety margins).
As stressed throughout this paper, COCOMIECOM are functional models that
make very few assumptions about internal cognitive structures and processes. For
present purposes, these models provide the starting point for a general conceptual
framework, based on which more specific models can be developed. Such models
should include both functional models as well as more detailed structural models
of the cognitive/neural mechanisms underlying behavioural effects of driver sup-
port functions (see, e.g. Victor, 2005 for a comprehensive review of models of the
latter type). Hence, functional and structural cognitive models should be viewed
as complementary and useful for different purposes. In addition to the further de-
velopment of more detailed driver models, the proposed conceptual framework is
a suitable starting point for framing hypotheses for empirical work on behavioural
effects of driving support functions. It is also useful as a common language in
industrial development of driver support functions, which helps maintaining the
focus on the purpose of the functions (e.g. how they should change driver be-
haviour), rather than their technological implementation (e.g. what type of sensors
are used).

References
Abelson, R.P. (1963). Computer simulation of "hot" cognition. In S.S. Tomkins & S Messick
(Eds.). Computer simulation ofpersonality. John Wiley and Sons, New York.
Alexander, G.J. and Lunenfeld, H. (1986). Driver expectancy in highway design and traffic
operations. US Department of Transportation. Report No: FHWA-TO-86-1.
Amditis, A., Polychronopoulos, A., Belotti, F. and Montanari, R. (2002). Strategy plan
definition for the management of the information flow through an HMI unit inside a car.
e-Safety Conference Proceedings, Lyon.
Bloch, H., Chemama, R., Gallo, A., Leconte, P., Le Ny, 1.-F. and Postel, 1. et aI., (Eds.)
(1999). Grand Dictionnaire de La Psychologie. Larousse, Paris.
Brookhuis, K.A., de Vries, G. and de Ward, D. (1991). The effects of mobile telephoning
on driving performance. Accident Analysis and Prevention, 23(4), 309-316.
Brostrom, R., Engstrom, 1., Agnvall, A. and Markkula, G. (2006). Towards the next gener-
ation intelligent driver information system (lDIS): The Volvo Cars Interaction Manager
concept. In Proceedings of the 2006 ITS World Congress, London.
Cacciabue, P.C. and Hollnagel, E. (2005). Modelling driving performance: A review of
criteria, variables and parameters. In L. Macchi, C. Re and P.C. Cacciabue (Eds.). Pro-
ceedings of the International Workshop on Modelling Driver Behaviour in Automotive
Environments. Ispra, Italy.
82 Engstrom and Hollnagel

Commission of the European Communities (2000). Annex to Commission Recommenda-

tion of 21 December 1999 on safe and efficient in-vehicle information and communication
systems: A European statement of principles on human machine interface. Official Jour-
nal ofthe European Communities L19, 25.1.2000, p. 64 ff, Brussels, Belgium: European
Union.
Damasio,A. (1994). Descarte'serror: Emotion, reason, and the human brain. G.P. Putnam's
and Sons, New York.
Dennett, D.C. (1987). The intentional stance. MIT Press, Cambridge, MA.
de Waard, D. (1996). The measurement ofdrivers' mental workload. ISBN 90-6807-308-7.
Traffic Research Centre. University of Groningen.
Dingus, T.A. (1995). Moving from measures of performance to measures of effectiveness
in the safety evaluation of ITS products or demonstrations. Paper presented at the safety
evaluation workshop. University of Iowa.
Donges, E. (1978). A two-level model of driver steering behaviour. Human Factors, 20(6),
691-707.
Engstrom, 1., Arfwidsson, 1., Amditis, A., Andreone, L., Bengler, K. and Cacciabue, P.C.
et aI., (2004). Meeting the challenges of future automotive HMI design: Overview of the
AIDE integrated project. In Proceedings, ITS in Europe. Budapest.
Engstrom, 1., Johansson, E. and Ostlund, 1. (2005a). Effects of visual and cognitive load in
real and simulated motorway driving. Transportation Research Part F, 8,97-120.
Engstrom, 1., Aberg, N., Johansson, E. and Hammarback, 1. (2005b). Comparison be-
tween visual and tactile signal detection tasks applied to the safety assessment of
in-vehicle information systems. In Proceedings of the Third International Driving Sym-
posium on Human Factors in Driver Assessment, Training and Vehicle Design. Rockport,
Maine.
Fosser, S., Saetermo, I.F. and Sagberg, F. (1997). An investigation of behavioural adaptation
to airbags and antilock brakes among taxi drivers. AccidentAnalyis and Prevention, 29(3).
Fuller, R.G.C. (1984). A conceptualisation of driving behaviour as threat avoidance. Er-
gonomics, 27, 1139-1155.
Fuller, R.C.G. (2005). Towards a general theory of driver behaviour. Accident Analysis and
Prevention, 37,461-472.
Gibson, J.1. (1979). The ecological approach to visual perception. Houghton Mifflin,
Boston.
Gibson, J.1. and Crooks, L.E. (1938). A theoretical field-analysis of automobile-driving.
The American Journal of Psychology, 51(3), 453-471.
Godthelp, H., Milgram, P.and Blaauw, J. (1984). The development of a time-related measure
to describe driving strategy. Human Factors, 26(3), 257-268.
Greenberg, 1., Tijerina, L., Curry, R., Artz, B., Cathey, L. and Grant, P. et al. (2003). Evalua-
tion of driver distraction using an event detection paradigm. Journal ofthe Transportation
Research Board, 1843. TRB, Washington DC.
Hale, A.R., Stoop, J. and Hommels, J. (1990). Human error models as predictors of accident
scenarios for designers in road transport systems. Ergonomics, 33, 1377-1388.
Harbluk, 1.L. and Noy, Y.I. (2002). The impact of cognitive distraction on visual behaviour
and vehicle control. Report no. TP 13889E. Transport Canada, Ontario.
Hollnagel, E. (1993). Human reliability analysis: Context and control. Academic Press,
London.
Hollnagel, E. (1998). Cognitive reliability and error analysis method. Elsevier Science,
Oxford, UK.
Hollnagel, E. (2004). Barriers and accident prevention. Ashgate Publishing Limited, Alder-
shot.
 4. A General Conceptual Framework for Modelling Behavioural Effects 83

Hollnagel, E. (2006). Outline of a function-centred approach to joint driver-vehicle system

design. Cognition, Technology & Work (in print).
Hollnagel, E., Nabo, A. and Lau, I. (2003). A systemic model for driver-in-control. Pro-
ceedings of the 2nd International Driving Symposium on Human Factors in Driver As-
sessment, Training, and Vehicle Design., Park City, UT. Public Policy Center, University
of Iowa.
Hollnagel, E. and Woods, D.D. (2005). Joint cognitive systems: Foundations of cognitive
systems engineering. Taylor and Francis/CRC Press, Boca Raton, FL.
Horrey, W.J. and Wickens, C.D. (2004). The impact of cell phone conversations on driv-
ing: A meta-analytic approach. Technical Report AHFD-04-2/GM-04-1, General Motors
Cooperation, Warren, MI.
Jamson, A.H. and Merat, N. (2005). Surrogate in-vehicle information systems and driver
behaviour: Effects of visual and cognitive load in simulated rural driving. Transportation
Research Part F, 8, 79-96.
Kahneman, D. (1973). Attention and effort. Prentice Hall, Englewood Cliffs, N1.
Lee, D.N. (1976). A theory of visual control of braking based on information about time-
to-collision. Perception, 5, 437-459.
Marken, R.S. (1986). Perceptual organisation of behaviour: A hierarchical control model
of coordinated action. Journal ofExperimental Psychology: Human Perception and Per-
formance, 12,267-276.
Markkula, G. and Engstrom, 1. (2006). A steering wheel reversal rate metric for assessing
effects ofvisual and cognitive secondary task load. In Proceedings ofthe 2006 ITS World
Congress, London.
McRuer, D.T., Allen, R.W., Weir, D.H. and Klein, R.H. (1977). New results in driver steering
control models. Human Factors, 19,381-397.
Michon, J.A. (1985). A critical review of driver behaviour models: What do we know? What
should we do? In Evans L.A. and Schwing R.C. (Eds.). Human Behaviour and Traffic
Safety. Plenum Press, NY, pp. 487-525.
Minderhoud, M.M. and Bovy, P.H.L. (2001). Extended time-to-collision measures for road
traffic safety assessment. Accident Analysis and Prevention, 33, 89-97.
Moray, N. (1967). Where is attention limited? A survey and a model. Acta Psychologica,
27,84-92.
OECD (1990). Behavioural adaptations to changes in the road transport system. Report
Prepared by an OECD Expert Group. Road Transport Research Programme.
Neale, V.L.,Dingus, T.A., Klauer, S.G., Sudweeks, 1. and Goodman, M. (2005). An overview
of the 1DO-car naturalistic study and findings. Paper presented at the 19th International
Technical Conference on Enhanced Safety of Vehicles (ESV). Washington DC, June 6-9.
Neisser, U. (1976). Cognition and reality. Freeman, San Francisco.
Nilsson, R. (2001). Safety margins in the driver. PhD Thesis. Uppsala University, Sweden.
Ostlund, 1., Nilsson, L., Carsten, 0., Merat, N., Jamson, H. and Jamson, S. et aI., (2004).
Deliverable 2 - HMI and Safety-Related Driver Performance (No. GRD1/2000/25361
S12.319626). Human Machine Interface and the Safety of Traffic in Europe (HASTE)
Project.
Powers, W.T. (1998). Making sense ofbehavior-the meaning ofcontrol. Benchmark, New
Canaan, CT.
Ranney, T. (1994). Models of driving behaviour. A review of their evolution. Accident
Analysis and Prevention, 26(6), 733-750.
Rasmussen,1. (1983). Skills, rules, and knowledge: Signals, signs, and symbols, and other
distinctions in human performance models. IEEE Transactions on Systems, Man, and
Cybernetics, SMC 13, 257-266.
84 Engstrom and Hollnagel

Recarte, M.A.and Nunes, L.M. (2003). Mental workload while driving: Effects on visual
search, discrimination, and decision making. Journal of Experimental Psychology, 9.
Saad, F., Hjalmdahl, M., Canas, 1., Alonso, M., Garayo, P. and Macchi, L. et aI., (2004).
Literature review of behavioural effects. Deliverable 1.2.1, AIDE Integrated Project,
Sub-project 1. IST-I-507674-IP.
SAVE-IT (2002). Safety vehicle( s) using adaptive interface technology (SAVE-IT) Program,
DTRS57-02-20003. US DOT, RSPSNolpe National Transportation Systems Center (Pub-
lic Release of Project Proposal), www.volpe.dot.gov/opsad/saveit/index.htmI.
Salvucci, D.D. (2001). Predicting the effects of in-car interface use on driver performance:
An integrated model approach. International Journal of Human-Computer Studies, 55,
85-107.
Shinar, D. (1993). Traffic safety and individual differences in drivers' attention and infor-
mation processing capacity. Alcohol, Drugs and Driving, 9, 219-237.
Smiley, A. (2000). Behavioural adaptation. Safety and Transportation Systems Research
Record, 1724, 47-51.
Simon, H.A. (1955). A behavioural model of rational choice. The Quarterly Journal of
Economics, LXIX, 99-118.
Summala, H. (1985). Modeling driver behavior: A pessimistic prediction? In Evans L,A"
Schwing R.C. (Eds.). Human Behaviour and Traffic Safety. Plenum Press, New York,
pp.487-525.
Summala, H. (1988). Risk control is not risk adjustment: The zero-risk theory of driver
behaviour and its implications. Ergonomics, 31(4),491-506.
van der Horst, R. (1990). A time based analysis of road user behaviour in normal and
critical encounters. PhD Thesis. Delft University of Technology, The Netherlands.
van der Horst, R., Godthelp, H. (1989). Measuring road user behaviour with an instrumented
car and an outside-the-vehicle video observation technique. Transportation Research
Record # 1213, Washington DC: Transportation Research Board, 72-81.
Weir, D.H., McRuer, T.M. (1968). A theory of driver steering control of motor vehicles.
Highway Research Record, 247, 7-39.
Wickens, C.D. (2002). Multiple resources and performance prediction. Theoretical Issues
in Ergonomics Science, 3(2), 159-177.
Victor, T.W. (2005). Keeping eye and mind on the road. PhD Thesis. Uppsala University,
Sweden.
Victor, T.W., Harbluk, 1.L. and Engstrom, 1. (2005). Sensitivity of eye-movement measures
to in-vehicle task difficulty. Transportation Research Part F, 8, 167-190.
Wilde, GJ.S. (1982). The theory of risk homeostasis: Implications for traffic safety and
health. Risk Analysis, 2, 209-225.
5
Modelling the Driver in Control
BJORN PETERS AND LENA NILSSON

5.1 Introduction

Modelling driver behaviour can be done with different purposes. One obvious aim
could be to provide a model of what drivers actually do and explain observed be-
haviour with or without support systems. Other objectives can be to discriminate
between safe and unsafe driving. Furthermore, a model can be used to imple-
ment a dynamic real-time control of driver support functions (see Chapter 6). The
following text is written in line with the first objective.
Safe driving requires that the driver is in control of the vehicle and the driving
situation. Driver support systems aim to help the driver and facilitate driving in
one way or the other. Support systems should be a response to identified existing
or potential problems related to the driving task. New technology like ADAS and
IVIS can facilitate driving but it could, on the other hand, make driving even
more complex and demanding. Solving one problem could be done at the cost of
introducing another. Thus, we need to understand driving task demands, driver
behaviour and how support systems can influence the driver's control and safety.
Here a broad overview of different theoretical approaches is given with focus
on control and safety. The use of a cognitive systems engineering approach is
advocated to investigate pros and cons of driver support systems. In the end, an
example is given on how findings from an experiment with a joystick controlled
car can be interpreted.

5.2 A Cognitive View of Driving

Driving is one of the most complex and safety critical everyday tasks in modern
society (Groeger, 2000). Driving a car is complex in the sense that it requires
the driver to employ a wide range of abilities in order to interact with a complex
environment and to manage the driving task demands. Driving is dynamic as the
demands can change back and forth from very low to extremely high, sometimes
within fractions of a second. When demands are high, driving is carried out in a

85
86 Peters, VTI and Nilsson, VTI

force-paced fashion; while as the demands are low, it can be performed in a more
self-paced manner. Normal driving can be considered as a cognitively motivated
and controlled task. Cognition is here used in the pragmatic sense as defined by
Neisser (1976), that is, cognition in context. A motive for this cognitive stance
can be found in Michon's (1985) visionary talk on driver behaviour modelling
' . .. the distinctly hierarchical cognitive structure of human behaviour in the traffic
environment . . . '. However, a cognitive approach does not imply that the driver's
perceptual and psychomotor abilities are to be neglected. Thus, driving can be
viewed as a cognitive task of control in a context perceived through the senses and
manipulated with control actions based on unconscious (automated) or conscious
decisions. This cognitive approach has been applied , for example , in adaptive
control models, control theory and cognitive systems engineering in order to model
driver behaviour.

5.3 Human Abilities

The human controller can be described in functional terms . In order to carry out
the driving task, the driver utilises three, tentatively different, functional abilities:
Cognitive, perceptual and motor abilities (see Fig. 5.1) . Cognitive abilities include,
for example, memory, decision, attention and supervision. Perceptual abilities can
be, for example, visual, auditory, tactile and proprioceptive. Finally, motor abilities
relate to physical dimension, motion and force , for example, reach, force and
endurance. These human abilities can be developed into very efficient functions
by train ing and experience. The abilities should not be considered as separate
entities but as highly interactive and even more so as they develop into skilled
behaviour. This functional description of the driver can be utilised to depict the
driver from a cognitive systems engineering view.

Cognitive
* mem OlY

21'
* attent on
Perceptual * decision
* vision
* hearing
* touch ~
* proprioception ----v
Physical
* size
* reach
*force ~
* endurance ~
FIGURE 5.1. Humanabilities em-
ployed by the driver.
 5. Modelling the Driver in Control 87

TABLE 5.1. Classification of driver behaviour models (adapted

from Michon, 1985).
Taxonomic Functional
Behavioural (input-output) Taskanalyses Adaptive control models
Internal state (psychological) Trait models Motivational models

5.4 Classifying Driver Behaviour Models

Several theories and models of driver behaviour have been presented, applied, anal-
ysed, criticised and abandoned. So far, no all-purpose, generic, comprehensive and
verifiable model of driver behaviour has been presented (Ranney, 1994). Differ-
ent models often emphasise specific and sometimes different aspects of driver
behaviour, for example, accident causation, education and training, behavioural
adaptation. Instead of searching for the ultimate model, it seems more feasible to
use a generic framework in order to get a structured view of existing driver be-
haviour models. Two different modelling approaches can be distinguished. The first
is to model what drivers actually do when driving. The second is to describe what
the drivers should do by modelling the driving task itself. The first can be called a
behavioural approach and the second a normative approach. Michon (1985) pro-
posed a generic classification of driver behaviour models, using a two-dimensional
classification (see Table 5.1). Firstly, he distinguished between behavioural or
input-output oriented models and internal state or psychological models. Sec-
ondly, he differentiated between functional and taxonomic models. Taxonomic
models are inventories of isolated facts, while functional models specify compo-
nents of driver behaviour and their dynamic relations. In this way he distinguished
between four types of models: task analysis models, trait models, adaptive control
models and motivational models. Task analysis models decompose driving into
tasks and subtasks and relate them to driver requirements and abilities. Trait mod-
els are based on the idea that it is possible to identify the accident-prone driver
with the use of well-designed tests. Functional models differ from taxonomic in
that they connect model components in order to consider the dynamics of driver
behaviour, for example, hierarchical structures. Adaptive control models apply
functional approach to capture behavioural changes. Finally, motivational models
consider internal states as attitudes, subjective risk and insight as controlling fac-
tors. The models that will be discussed in the following can be mostly categorised
as adaptive control models. For a discussion on other types of models, see Carsten
(2005, Chapter 6).

5.5 Hierarchical Control Models

Most models can offer only post hoc explanations of observed behaviour
or possibly explain aggregated accident data (Rumar, 1988). Early cognitive
models focused on information processing, and driving was considered as a
88 Peters, VTI and Nilsson, VTI

problem-solving task. Accidents were attributed to incorrect information process-

ing, and Rumar concluded that there was an urgent need to develop models and
hypothesis that can predict actual driver behaviour. Michon (1985) claimed that
the lack of progress emerged from the failure to consider results from cognitive
psychology. Later models incorporate developments from cognitive psychology,
for example, hierarchical control structures and automaticity (Ranney, 1994).
Rumar (1988) pointed out that an important behavioural uniqueness of driving
can be its combination of consciously controlled (cognitive) and unconsciously,
automatic (perceptual) behaviour. Hierarchical control models became more
accepted when Michon (1985) advocated the idea that such an approach would
resolve some of the identified shortcomings with earlier models. However, this
was not a new idea, for example, Allen et al. (1971), McRuer et al. (1977) and
Janssen (1979) described driving as a hierarchical structured task with strategic,
tactical and operational components demanding different levels of driver control
much earlier. At the strategic level, the driver is concerned with tasks such as
planning the journey, selecting the mode of transport and choosing a route. At the
manoeuvring level, the tasks concerned include overtaking, giving way to other
vehicles and obeying traffic rules. At the control level, the driver is concerned
with controlling the vehicle, for example, controlling speed, following the road
and quite simply keeping the car on the road. The model assumes an interaction
between the three levels in which goals and criteria are defined at a higher level
and the outcome of lower levels modifies goals at a higher level. Finally, Michon
(1985) meant that a comprehensive model of driver behaviour should not just
identify different levels of control but also provide a control structure that enables
control to shift from one level to another in a timely manner.
The hierarchical structure of the driving task can be matched to actual driver
behaviour. Human control structures are highly flexible and highly dependent on
practice and experience. These structural aspects of human performance were ad-
dressed in the hierarchical skill-, rule- and knowledge-based behaviour (SRK)
model developed by Rasmussen (1986). Rasmussen identified a number of cog-
nitive functional elements organised in a three-layered structure. The model dis-
criminates between skill-based, rule-based and knowledge-based controls. The
main issue in Rasmussen's framework is the hierarchical nature of human control
replacing the serial model used in early models of human information process-
ing. The SRK model has been extensively used to model driver behaviour. The
hierarchical control model takes into account both the hierarchical structure of the
driving task and the driver behaviour by combing the two frameworks mentioned
above (Ranney, 1994). Thus, Michon (1985) compiled a two-dimensional matrix
that has been used to explain driver behaviour and to identify driver support needs
(Michon, 1993; Ranney, 1994; Nilsson et aI., 2001) with driving task demand and
driver behaviour as the two dimensions (see Fig. 5.2). New driver support systems
might force the driver to learn new skills and forget old skills. As a consequence,
the driver might have to shift from skill-based behaviour to knowledge-based be-
haviour in order change to new skill-based behaviour, for example, learning to use
an ADAS type of system like ABS.
 5. Modelling the Driver in Control 89

Skill based

Strategical

Driving
Task Tactical
Demands I - - - -I -- - - -+-____

Control

FiGURE 5.2. IVIS and ADAS can influence both driving task demands and driver behaviour
(from Nilsson et al., 2001) .

With the hierarchical structuring of both task and behaviour, it becomes evident
that time is an aspect of driving that should be considered. Time constraints are im-
plicitly different for the three task levels, even if time is not specifically addressed
in the hierarchical control model. The following approximate time frames apply:
lOs or more for tasks at strategic level, between I and lOs at tactical level and less
than I s at control level. Finally, it should be noted that more complicated driver
tasks, for example, performing route planning while driving or mobile phoning
during driving may require knowledge-, rule- and skill-based actions in combina-
tion . It is not simply the time requirements that distinguish the different levels of
tasks, but also the requirements of attention, workload and the consequences of
mistakes, etc. The hierarchical control models are functional models, which do not
specifically include motivational aspects and should be classified as adaptive con-
trol models . Furthermore, they do not explicitly consider the context, for example,
controller's influence on the system to be controlled (Hollnagel, 2000) .

5.6 Control Theory

Control theory or cybernetics is a general theory aimed at understanding self-
regulating systems (Wiener, 1954; Ashby, 1956; Carver and Schreier, 1982). The
systems view approach applied in cybernetics distinguishes itself from the more
traditional analytic approach by emphasising the interactions and connectedness
of the different components of a system . The basic control theory idea is simply
based on a stimulus-response loop . The system current status is compared to a
reference value and the deviation between system current status and reference
determine the control action. However, the definitions of system boundaries and
the relation between cause and effect are different from, for example, information-
processing models . Control theory has been used to describe and study a wide
range of systems from ' pure' technical to biological and social systems . Adaptive
control models are also based on control theory.
90 Peters, VTI and Nilsson, VTI

Stimulus

Environment

-------------: Response/Stimulusr -------------

Action I _;Sensation
--[----------: EnvIronment:---------- ----
I--------------'"j I---------------J
1 1 1 I

1
I
Decision ~
1 1
Perception .I
~ J ~-------------..!

FIGURE 5.3. Open loop - closed system versus closed loop - open system adapted from
Jagacinski and Flach (2003).

A system is an abstract construct, used to identify the focus of interest (Jagacin-

ski and Flach, 2003). The system is also sometimes contrasted to the environment.
A system typically refers to the phenomenon of interest and the environment
to everything else. When there is a sharp boundary between the system and the
environment, the system is considered as a closed system. The environment does
not influence the system and can be disregarded. In a closed system, control is
well defined and deterministic. Such a closed systems view is applied in human
information-processing (HIP) models (e.g., Wickens, 1992) when the controller
(e.g. driver) is considered as a closed system and the controlled object (e.g. car)
belongs to the environment. The rationale for this approach was to divide human
behaviour into isolated entities (e.g., sensation, perception, decision and action)
which could be studied independently (see Fig. 5.3, top). As can be seen, it is a
closed system (detached from the environment) and the control is an open-loop;
that is, there is no feedback depicted from the controller's output to system input.
However, the demarcation between a system and the environment is often not
that sharp. Or rather there are phenomena that cannot be explained with a closed
systems view. This is specifically true when concerned with behavioural science
where the relationship between, for example, a controller and the environment
seems to be of prime importance in order to understand complex phenomena like
behavioural adaptation. When the demarcation between a system and the environ-
ment is diffuse, the system can be considered an open system. All socio-technical
systems can be viewed as open systems (Flach, 1999). The primary difference with
other SR models is that control theory considers the operator's influence on the
controlled object and the system boundaries are defined so that both the operator
and the controlled object are included. This open system view is more consistent
with an ecological view (Gibson, 1966). System boundaries can be very tight to
all embracing depending on the focus. This open system view can be depicted by
closing the loop and position the environment in the centre (see Fig. 5.3, bottom).
 5. Modelling the Driver in Control 91

By closing the loop, the cause - effect relation between stimulus and response
loses its meaning. In a closed-loop system the stimulus and response are tightly
linked and there is no clear distinction between what is cause and what is effect.
This restructuring was done to illustrate that the stimuli are as much determined
by the actions as the actions are determined by the stimuli (Jagacinski and Flach,
2003). In this way we have a closed-loop open system, as the environment is
included and the interdependencies between the controller and the environment
are considered. This view can be useful when modelling driver behaviour and, in
particular, driver behaviour adaptation and driver support systems.
Cognitive Systems Engineering (CSE) put cognition in context by applying an
overall systems view. CSE decomposes complex tasks along two dimensions ab-
stract - concrete and whole - part (Jagacinski and Flach, 2003). In this way the
focus is shifted from internal functions of either humans or machines to the ex-
ternal function making up a joint cognitive system including both the controller
and the controlled system (Hollnagel and Woods, 2005). The driver will be viewed
as a controlling system and the car as a technical system to be controlled. In this
perspective the car will constitute the primary context for the driver. As more of the
context is included, the system will expand as a multi -layered functional descrip-
tion. The system boundaries have to be defined according to the purpose of the
analysis. The two most important concepts in Control Theory are the open/closed
system and the open/closed loop control. Jagacinski and Flach (2003), among oth-
ers (e.g. Hollnagel and Woods, 2005; Hollnage12002), have further developed the
concepts of control theory. Jagacinski and Flach have also provided some quan-
titative tools based on control theory that can be used to capture and understand
also qualitative aspects of human performance, for example, driving behaviour and
driver support systems.

5.7 Adaptive Control Models

Adaptive control models are concerned with issues of how the driver adapts his or
her control to the characteristics of the system to be controlled (driver-vehicle-
environment). Two categories of adaptive control models can be distinguished
servo-control models addressing continuous tracking and information flow control
models addressing discrete decision making (Michon, 1985). However, both views
can be used in a complementary manner to understand some aspects of manual
control. For example, firing the motor neurons is a set of discrete all-or-none
response. However, the motion of an arm depends on the integration over many
neurons and can best be described as continuous at a coarse time scale. Early
servo-control models were mostly used to study compensatory (feedback) steering
control performance on roads with varying curvature and for evasive manoeuvres.
McRuer et al. (1977) questioned this way of modelling steering behaviour, as it
did not consider the anticipatory control that seems to be a behaviour needed in
order to understand skilled driving behaviour.
92 Peters, VTI and Nilsson, VTI

External _ _-----,
disturbances
Lateral
position

Heading
angle

Desired
path

FIGURE 5.4. A three-level servo-control model of steering (from McRuer et al., 1977).

McRuer et al. (1977) proposed a three-level servo-control model of drivers'

steering behaviour. First of all, they described driving as consisting of a hierarchy
of navigation, guidance and control phases conducted simultaneously with visual
search, recognition and monitoring operations. They also distinguished between
closed-loop (compensatory) control and open-loop (anticipatory) control. Com-
pensatory steering was described as two feedback loops (see Fig. 5.4). Firstly,
the lateral position is fed back and compared to the desired path, and if there
is a deviation it will result in an error-correcting action, which is compared to
current heading angle and, if needed, a steering wheel correction will be made.
The perceived road curvature derived from visual input guides the pursuit con-
trol. Secondly, pursuit control is an open-loop feed-forward control element that
permits the driver to follow the anticipated road curvature. An interesting third
concept is the precognitive control that in practice is a first phase of dual-mode
control, that is, both open- and closed-loop controls. Precognitive control consists
of previously learned control actions, which are triggered by situation and vehicle
motion but work as pure open-loop control.
In view of this model, steering can be considered in terms of output as a position,
velocity or acceleration control system (Jagacinski and Flach, 2003). If tyre angle
is considered to be the output, then the steering system can be approximated as a
position control system with a gain given by the steering linkage. While, if heading
angle is viewed as the output, then steering can be considered a rate control system.
In this case the gain is proportional to the velocity of the front wheels. Finally,
if lateral position is considered the output, then steering should be viewed as an
acceleration control system. If so, then the effective gain between steering wheel
angle and lateral position is proportional to the square of the velocity. This shows
that lateral and longitudinal control of the car is not independent but very much
entwined. For a further discussion on control theory modelling, see also Wier and
Chao (2005, Chapters 16 and 17).
Even if McRuer and his colleagues (1977) described driving as a hierarchical
task, they concentrated very much on steering control. They largely considered
driving as a closed system and disregarded the environment except road geometry.
Michon (1985) cited Reid (1983), who concluded that the servo-control model
 5. Modelling the Driver in Control 93

described above cannot successfully cope with driver tasks other than follow-
ing straight and smoothly curved roads. The model needs to be better integrated
with the guiding visual environment as described by, for example, Gibson (1966).
Michon concluded that 'The two fields - perception and vehicle control - are
still lacking a theoretical integration. Combining them would constitute a major
breakthrough, ... ' .

5.8 Cognition in Control

Most of the models discussed so far are basically mechanistic, as they do not
recognise the need of higher order cognitive abilities. The interaction between the
driver and the environment is not explicitly included in the model but rather seems
to be an implicit presumption. A useful model needs to be better connected with
the context. The lack of context in cognition was addressed by Neisser. Neisser
(1976) criticised the concept of direct perception (Gibson, 1966) by stating that
to see is not just to perceive but also to interpret and understand on a conscious
level. Neisser (1976) proposed a cognitively driven model of perception called
the perceptual cycle, which includes the interaction between the observer and the
environment. He introduced what he called anticipatory schemata that prepare and
control our perception. Neisser (1976) further meant that human control works in a
way similar to the perceptual cycle. Thus, to control a system, the controller has to
have a model of the system to be controlled. The importance of a control-guiding
model can also be understood in the light of 'The law of requisite variety' (Ashby,
1956), which states in principle that the variety of the controller should match the
variety of the system to be controlled. Thus, the controller's understanding of the
system that is being controlled will determine the actual control actions. In other
words the driver's mental model of the vehicle, other drivers, road condition, etc.,
will determine the driver's control behaviour.
Hollnagel and Woods (2005) described a cyclical model of control, the basis
of on the principles of Neisser's perceptual cycle. Hollnagel's cyclical control
model was used as the basis for the Contextual Control Model, which describes in
general terms how performance depends on perceiving feedback events, interpret-
ing and modification of current understanding, selection and execution of actions.
Driver control behaviour can be described as shown in Fig. 5.5, which is based on
Hollnagel's cyclical model of control (Chapter 4).
The control cycle is divided into three phases: perception, decision and action.
The control cycle is cognitively initiated by the driver depicted with the arrow
coming out of the driver's head. The driver's mental model of the system to be
controlled and the environment will guide the search for information during the
perceptual phase. The perceived situation is compared to a reference value defined
by the driver. The comparison is followed by a cognitive phase. During this phase
a decision will be made on the basis of the difference between reference value and
current situation, that is, the error. The aim will usually be to minimise the error.
This cognitive phase is followed by an action phase during which an appropriate
94 Peters, VII and Nilsson, VTI

FIGURE 5.5. The cognitive control

cycle adapted from Hollnagel' s
(Hollnagel and Woods, 2005)
d Externa l events

r'ITeA~
contextual control model.

action is selected and carried out. This action influences the environment depicted
by the outward arrow. Once the action is carried out, the driver searches and per-
ceives the effect of the action together with possible external events and the circle
is closed. The result of the action phase is also fed back to the driver and will
change the driver's mental model of the control loop and the system under con-
trol. In other words it is previous experience and outcome of the driver's actions
on the controlled system that will form and develop his or her mental models.
The three phases are described as three distinct entitie s but in reality the phases
might be overlapping and not separated as might appear from the figure. How-
ever, in principle the three phases are different in character. This model of driver
control provide s a foundation to capture the dynamics in driving, for example ,
compensatory closed-loop and anticipatory open-loop driving .
In relation to the discussed control model it can be interesting to consider the
dual control problem: The driver should both determin e the system status and at
the same time control it (Jagacinski and Flach, 2003). The driver occasionally has
to get out of control in order to maintain control. This was described by Weinberg
and Weinberg (1979) as the fundamental regulator paradox: The lesson is easiest
to see in terms ofan expe rience common to anyon e who has ever driven on an icy
road. The driver is trying to keep the carfrom skidding. Toknow how much steering
is required, she must have some inkling of the road' s slickness. But ifshe succeeds
in completely preventing skids, she has no idea how slippery the road really is.
Good drivers. expe rienced Oil icy roads, will intentionally test the steering from
time to time by 'jiggling ' to cause a small amount of skidding. By this technique
they intentionally sacrifice the perfect regulation they know they cannot attain in
any case. III return. they receive information that will enable them to do a more
reliable. though less pe rfect job. Unintentional skidding will, of course, provide
sufficient information to the driver on how to adapt the driving behaviour in order
to overcome the slippery road condition.
 5. Modelling the Driver in Control 95

5.9 Goals for Control

The control theory model discussed above assumes that there is a goal or a reference
value, which is used to determine appropriate actions. Different goals for safe
driving behaviour have been proposed, for example, zero risk, threat avoidance,
safety margins, response to risk (Naatanen and Summala, 1974; Fuller, 1984;
Summala, 1985, 1988; Groeger, 2000). However, one of the first models aiming
to describe safe (normal) driving behaviour was presented by Gibson and Crooks
(1938). They meant that driving is a task that is carried out in time and space and
described driving as a task of controlling the car within the field of safe travel.
They defined the field of safe travel as 'an indefinite bounded field consisting, at
any moment, of the field of possible paths which the car may take unimpeded' . It is
an imaginary dynamic area in front of the vehicle with a shape of an outstretched
tongue (see Fig. 5.6). Obstacles in the terrain mainly determine the boundaries
in the field. Gibson and Crooks' model provides a foundation for the concept of
safety margins and a guiding mechanism for normal safe driving behaviour.
Driving within safety margins is a concept that has been explored by several
researchers. As an example, the concept of the longitudinal safety margins was
explored by van der Hulst (1999). Van der Hulst meant that experienced drivers
have a mental model of the driving task and expectations about what will occur
in which situation. As a consequence, the driver can effectively scan the traffic
environment. The mental model and the expectations will allow the driver to build
up and preserve situation awareness and also to take advantage of a precognitive
control as described by McRuer et al. (1977). Furthermore, driving is a task that
can be carried out in many ways and will provide opportunities for behaviour

I
,I
I
I
I

,/
,./
,,/'.
.'

FIGURE 5.6. The concept of field of safe travel (from Gibson and Crooks [1938] published
with permission of The American Journal of Psychology).
96 Peters, VTI and Nilsson, VTI

adaptation. The driver can choose from several driving strategies. This is a view that
can help to better understand the mechanism behind the anticipatory behaviour. Van
der Hulst (1999) also meant that driving is a task that allows for pace adjustments
by means of adjustments in speed and other safety margin. The driver's choice
of speed and safety margins will determine the time available to react to relevant
changes in the environment. Van der Hulst thereby connected Gibson and Crooks's
'field of safe travel' with the Ashby's 'law of requisite variety'. Summala (1985,
1988) proposed that safety margins could be operationally defined as distance-
or time-related measures like Time-to-line-crossing (TLC) and Time-to-collision
(TTC). The concept of safety margins can also be used to explain, at least partly,
accident causation. In-depth accident studies have shown that late detection is a
very common explanation given for collisions (Rumar, 1988). Late detection can
be described as violation of safety margins. Thus, time is critical for safe driving,
which will be discussed later.
However, driver behaviour is most likely determined by a set of concurrent
goals. The goals might also shift during a drive. Given the hierarchical structure
of the driving task and driver behaviour as described above, it seems likely that
the driver applies different goals for different levels of control. Thus, the driver
can at the same time drive to reach a destination in time, stick to the traffic rules
and avoid accidents. Goals for driving behaviour can also differ between drivers
and situations. Furthermore, goals for driving can be extended to incorporate, for
example, goals for life, skill for living, sensation seeking, pleasure, etc. (Hatakka
et aI., 2002). Thus, it seems likely that the driver has to find a balance between
different goals. The idea of balancing between different goals was applied by
Wilde (1982) in the risk homeostasis theory. Homeostasis is originally a term
used to describe a complex mechanism for maintaining metabolic equilibrium in
biological systems. Wilde meant that driver behaviour is guided by a target risk
level that is determined by a combination of subjective and objective risk. An
implication of Wilde's approach is that all actions taken to improve safety will
be neutralised by the driver - at least on an aggregated level. The consequences
of Wilde's theory led to considerable controversy and the theory has been re-
jected by several researchers (e.g. McKenna, 1982; Michon, 1985; Sanders and
McCormick, 1993). According to Fuller (2005, Paper 10), Wilde based his the-
ory on a misinterpretation of empirical findings made by Taylor (1964). Fuller
proposed instead that task difficulty homeostasis, which describes the dynamic
interaction between driving task demands and driver capability, could be used
as key-sub goals to describe driver behaviour. Fuller argues that the task diffi-
culty homeostasis overcomes some problems with the safety margin concept and
conforms to the hierarchical structuring of the driving task and driver behaviour
and risk homeostasis theory can be viewed as a special case. Other human be-
haviour researchers has shown how homeostasis can be used to understand how,
for example, emotions can influence behaviour (Damasio, 1999; see Vaa, 2005,
Paper 12). Thus, it seems likely that homeostasis is a mechanism that could be
explored further to understand how different goals interact and determine driver
behaviour.
 5. Modelling the Driver in Control 97

5.10 Time and Time Again

Time considerations are crucial for the cognitive control cycle described above.
Hollnagel and Woods (2005) divided the control cycle in three different phases:
perception, decision and action. Time constraints can be incorporated into the
cognitive control cycle (see Fig. 5.7), slightly adapted from Hollnagel. Thus, speed,
road geometry, obstacles in the field of travel, sight conditions among a range of
other factors determine the time available for the control cycle. This time is labelled
Tu (usable time). The times needed to carry out the three phases (perceive, decide
and act) are labelled Tp , Td and Ta , respectively. In 'normal' driving Tp + Td + Ta
is less than Tu .
Depending on the situation, normal driving constitutes a combination of com-
pensatory and anticipatory control. The driver usually strives to balance between
compensatory and anticipatory driving. Anticipatory driving requires Tu to be
longer than (Tp + Td + Ta ) but as the time required comes close to what is avail-
able driving becomes more compensatory. If the total usable time is not sufficient,
performance will start to degrade. The control becomes more erroneous or slug-
gish and oscillatory (Jagacinski, 1977). Reducing speed is one way to gain time
and control. The model also depicts that the reason for deteriorated performance
can be attributed to prolonged perception, prolonged decision, prolonged action
or some combination of the three. In any case, the result will be that the used
time will be more than the usable time. All three phases are connected, meaning
that if one part requires less time than expected then there will be more time for
the remaining two phases. When traffic demands are low and the driver is ex-
perienced, evaluation, selection and even action require little time and there will
be plenty of time available. Time pressure is in this view a critical component
of driving behaviour when determining the driver's safety margins. Closely re-
lated to this is the concept of uncertainty. The driver will try to keep uncertainty
at an acceptable level to maintain pace control and safety margins (Godthelp,

external
events

Tu= total usable time

Tp=time needed to perceive
Td=time needed to decide
Ta=time needed to act

FIGURE 5.7. Time and control in the cyclic control model (adapted from Hollnagel, 2002).
98 Peters, VTI and Nilsson, VTI

Milgram and Blauw, 1984; Wierwille, 1993). The driver can apply an intermittent
sampling strategy to cope with lack of time for the control loop. However, this
strategy can, if maintained, increase the level of uncertainty, depending on the
situation (Lee, 2006). Increased uncertainty and lack of anticipation will make
drivers vulnerable to accidents. The time concept in the control cycle described
above can also be applied to driver support systems. If the driver's mental model
of a support system is incorrect or incomplete, this misunderstanding can pro-
long the time needed to complete the control loop. If the support system provides
delayed or contradictory feedback and the driver is not guided to the right ac-
tion, then the driver might eventually lose control. Lost control can also be due
to a situation when the driver requires a long time to perceive what the system is
doing.

5.11 Multiple Layers of Control

The driving task can be structured in hierarchical levels as discussed earlier. So
far we have mostly been concerned with low levels of control. To get a more
holistic view of driver behaviour, we need to expand the cognitive control loop to
incorporate more of the context and apply a joint systems view (driver-vehicle-
road-traffic etc.). The extended control model (ECOM) was developed by Holl-
nagel and Woods (2005) to consider aspects of joint system control. Hollnagel
distinguished four hierarchical layers of control: tracking, regulating, monitor-
ing and targeting, with targeting as top level and controlling at the bottom level.
These four layers correspond to the levels distinguished in a generic decision
model. Hollnagel pointed out that there is no absolute reference that can be used
to determine the number of layers needed for all cases. Rather, the purpose de-
termines the number of layers needed; for example, Powers (1998) identified 11
levels in his perceptual control theory. The layers in Hollnagel's model should
not be considered as distinctive but more as continuous and overlapping. The
control layers are connected in such a way that goals for control at lower layers
are determined at higher layers and feedback is provided from lower to higher
layers. The primary control of the car, that is, lateral and longitudinal motion of
the vehicle, involves all control layers even if tracking control is the most obvi-
ous. Furthermore, the driver has to perform several tasks - not just the primary
control of the car but also several additional tasks that are often carried out in par-
allel. Such additional tasks can be interaction with secondary car controls, driver
support systems and even nomad systems, for example, mobile phones, hand-
held computers, etc. Even if a driver support system can be aimed at a specific
control layer, it might very well affect control on other layers. For a more elabo-
rated description and application of the ECOM in relation to driver support, see
Paper 4.
Given that driving is considered as a cognitively motivated and controlled task,
it is possible to reorganise the four control layers in ECOM starting with the top
level of control targeting as an inner control loop and the monitoring, regulating
 5. Modelling the Driverin Control 99

FIGURE 5.8. An extended control

model adapted from Hollnagel's ex-
tendedcontrol model (Hollnagel and
Woods, 2005).

External events

and targeting loops as concentric circles with increasing diameters (see Fig. 5.8).
Thus, what Hollnagel described as the lowest level of control tracking will be the
outer circle representing the physical interaction with the interface to the vehicle 's
physical controls (e.g. steering wheel, pedals or various driver support systems).
Control goals are determined in the inner control loops and applied in the outer
control loops. That is to say that, for example in the targeting control loop the
goals are determined for the monitoring control loop. This flow is represented by
the outward-bound arrow-labelled goals at the top of Fig. 5.8. Feedback used to
modify and supervise the control is fed back from outer circles to inner control
circles represented by the in-bound arrow-labelled feedback at the top of Fig. 5.8.
The interaction between the driver and the physical environment is represented by
the two arrows-labelled external events and actions at the bottom of Fig. 5.8.
The driver can conduct an imaginary drive without physical interaction on the
targeting level. The driver can also, while driving , anticipate the possible outcome
of possible actions and let that determine the choice of action . However, even if the
cognitive skills are important, it is the lower level psychomotor skills that are the
foundation for safe driving (Ranney, 1997). Thus both cognitive and psychomotor
skills are needed for safe driving and furthermore the integrations and interaction
between the skills. The driver's mental model is, to a large extent, formed by the
experiences from driving - ' learning by doing' . Driver support can be provided
at different layers or levels of control and it seems likely that the ECOM can be
used to analyse and better under stand how different driver support systems can
influence driving behaviour.
100 Peters, VTI and Nilsson, VTI

5.12 Joystick Controlled Cars - An Example

Experiences from experiments with four-way joystick controlled cars for drivers
with severe disabilities will exemplify the application of the model discussed. The
joysticks were used to control speed (accelerating and braking) and steering and
thus replaced the standard controls (pedals and steering wheel). A manoeuvre test
on a closed track was performed with experienced drivers of joystick-controlled
cars (Ostlund and Peters, 1999). The test included the following three manoeuvres:
(1) firm and controlled braking on straight road, (2) firm and controlled braking
in a narrow curve, and (3) double-lane change. It was found that the test drivers
found it difficult to perform the straight road brake manoeuvre smoothly, which
was attributed to the lack of feedback. Braking in a curve was a manoeuvre, which
revealed problems of control interference between steering and braking control.
Finally, time lags in the steering control system caused problems, specifically when
performing the double lane-change manoeuvre.
The lack of tactile feedback when braking on a straight road was something that
mainly influenced the activities at the lowest level, tracking. The drivers were not
able to adjust the force applied on the brake lever in order to make a soft stop.
It can be speculated that the feedback they experienced, as whole-body g-forces
together with the visual cues, was not sufficient for the brake control function.
However, it is probably a delicate task to implement feedback forces to suit the
category of drivers who participated in the test, for example, drivers with sever
muscular dystrophy. Adaptation companies installing these systems often claim
that these drivers are extremely weak that active force feedback cannot be used.
However, this is probably not true because all manual control depends on some
form of tactile feedback. Whether it is technically feasible is another issue. In that
case it might be possible to investigate other sources of feedback, for example,
auditory feedback. However, such feedback will be artificial and not intuitive in
the same way as force feedback would probably be.
The interference problem observed during curve braking mainly affected the
activities on the regulating and monitoring levels. The drivers probably had a
mental model of how their joysticks worked in terms of steering and braking but it
was not sufficient to know what direction to move the lever in order to brake without
affecting steering control. They had to regulate and monitor their control actions
closely in order to adjust the joystick motion. This type of manoeuvre needs to be
carried out at least partly as anticipatory control. With increased experience the
drivers are likely to develop motor control schemata comparable to the precognitive
control proposed by McRuer et al. (1977). It was also observed that the drivers
compensated for performance decrements by driving slowly in the curve.
Finally, the time lag problem observed in the double-lane change had to be
handled as anticipatory control. The drivers initially had to plan (target) in advance
how to move the lever in order to compensate for the time lags and maintain control
in the double lane change. It seems very difficult to develop a mental model of
time lags and requires a lot of experience and training. It also turned out that the
double lane change manoeuvre was the most difficult task to carry out correctly
 5. Modelling the Driver in Control 101

(i.e., without hitting any cones). The joystick drivers performed worse than a group
of drivers matched with respect to age and driving experience. These drivers drove
a standard car. The joystick drivers hit more cones and produced higher lateral
acceleration forces despite driving their own individually adapted cars. All of the
identified problems with these joystick systems can be described and analysed in
terms of time-based safety margins.
Practical experiences have shown that it is a tedious task to learn to drive with this
type of control system (often implemented as a electro-hydraulic system). Thus, in
a follow-up simulator experiment an alternative joystick design was tested (Peters
and Ostlund, 2005). Time lags had been made similar to what is found in ordinary
car controls (steering wheel and pedals), steering and speed control had been made
more independent and active feedback was provided in the joystick lever. It was
found that the reduced time lag contributed substantially to make it easy to learn
to drive with this joystick. The separation of steering and speed control did not
as clearly improve performance but contributed somewhat to improved control.
The active feedback was not sufficiently tuned according to the individual drivers.
Thus, it was only drivers with unimpaired upper limb functions who could benefit
from the feedback.
In the example above it is obvious that the drivers were in need of support as
they could not drive a standard production car due to their physical impairment
(i.e., they were paralysed in their lower limbs and had impaired function in their
upper limbs). However, it can well serve as an example on how to understand
potential problems related to specifically ADAS and guide on how to resolve
them. What becomes obvious is that a driver in control depends on timely, sufficient
and intuitive feedback. Furthermore, integrated control functions with a high risk
of interference should be avoided. The human controller can often learn how to
compensate even badly designed support systems but a well-designed system will
both facilitate learning and ensure that the driver stays in control even in a critical
situation.

5.13 Summary and Conclusion

Driving task and driver behaviour modelling can be valuable to determine driver
support needs. Modelling can also be useful to determine whether the objectives
with a support system have been reached or to look for possible aversive conse-
quences. A driver support system can have different purposes and a range of dif-
ferent consequences for driver behaviour. However, the main focus in this Paper is
on driver support and safety. It has been argued that driving should be considered
a cognitive task of control in context. Furthermore, the hierarchical structure of
both the driving task and driver behaviour has to be considered. The principles of
control theory can be used to understand the driver as an active controller. Context
has to be included and system boundaries have to be determined. To understand
complex phenomena like behaviour adaptation, it was proposed to apply a closed
loop open system view. Thus, the cognitive control cycle that can be linked to
102 Peters, VTI and Nilsson, VTI

basic human abilities needed for the driving task was proposed as a useful model
of driver behaviour. Driver behaviour is also guided by a set of concurrent goals.
The homeostasis principle seems to be useful to understand the interrelationship
between different goals and driver behaviour. Human control seems to be very
sensitive to time. Time determines the margins for action. The cognitive control
cycle was described in terms of time. Finally the control cycle was expanded to
also consider hierarchical layers of control and the interaction with the system to
be controlled. The overall aim should be to ensure that the driver is in full control
even when driver support systems are installed. Well-designed support systems
are easier to learn and safer to use.

References
Allen, T.M., Lunenfeld, H. and Alexander, G.1. (1971). Driver information needs. Highway
Research Record, 366, 102-115.
Ashby, W.R. (1956). An introduction to cybernetics. Chapman and Hall Ltd., London.
Carsten, O. (2005). From driver models to modelling the driver: What do we really know
about the driver? In L. Macchi, C. Re and P.C. Cacciabue (Eds.). Modelling driver
behaviour in automotive environments. Office of Official Publiations of the European
Communities, Ispra, Italy.
Carver, C.S. and Schreier, M.F. (1982). Control theory: A useful conceptual framework for
personality - social, clinical and health psychology. Psychological Bulletin, 92, 111-135.
Damasio, A.R. (1999). The feeling of what happens: Body and emotion in the making of
consciousness. Harvest Book Harcourt Inc., San Diego, CA.
Engstrom, 1. and Hollnagel, E. (2005). Towards a conceptual framework for modelling
driver' interaction with in-vehicle systems. In L. Macchi, C. Re and P.C.Cacciabue (Eds.).
Modelling driver behaviour in automotive environments. Office of Official Publications
of the European Communities, Ispra, Italy.
Flach, 1.M. (1999). Beyond error: The language of coordination and stability. In P.A.
Hancock (Ed.). Handbook of perception and cognition: Human performance and er-
gonomics. Academic Press, New York.
Fuller, R.G.C. (1984). A conceptualisation of driving behaviour as threat avoidance. Er-
gonomics, 27, 1139-1155.
Fuller, R. (2005a). Control and affect: Motivational aspects of driver decision-making. In
L. Macchi, C. Re and P.C. Cacciabue (Eds.). Modelling driver behaviour in automotive
environments. Office of Official Publications of the European Communities, Ispra, Italy.
Fuller, R. (2005b). Towards a general theory of driver behaviour. Accident Analysis and
Prevention, 37,461-472.
Gibson, 1J. (1966). The senses considered as perceptual systems. Houghton Mifflin, Bostan.
Gibson, 1.1. and Crooks, L.E. (1938). A theoretical field-analysis of automobile-driving.
The American Journal of Psychology, 51,453-471.
Godthelp, H., Milgram, P. and Blaauw, GJ. (1984). The development of a time-related
measure to describe driving strategy. Human Factors, 26, 257-268.
Groeger, 1.A. (2000). Understanding driving -Applying cognitive psychology to a complex
everyday task. Psychology Press, Taylor and Francis, Hove, UK.
Hatakka, M., Kesikinen, E., Gregersen, N.P., Glad, A. and Hernetkoski, K. (2002). From con-
trol of vehicle to personal self-control; broadening the perspectives to driver education.
Transport Research Part F, Traffic Psychology and Behaviour, 201-215.
 5. Modelling the Driver in Control 103

Hollnage, E. and Woods, D.D. (2005). Joint cognitive systems: Foundations of cognitive
systems engineering. Taylor and Francis, Boca Raton, FL.
Hollnagel, E. (2002). Time and time again. Theoretical Issues in Ergonomics Science, 3,
143-158.
Jagacinski, R.I. (1977). A qualitative look at feedback control theory as a style of describing
behavior. Human Factors, 19(4),331-347.
Jagacinski, R.I. and Flach, 1.M. (2003). Control theory for humans-Quantitative ap-
proaches to modelling performance. Lawrence Erlbaum, Mahwah, NJ.
Janssen, W.H. (1979). Routeplanning en geleiding: Een litteratuurstudie. Institute for per-
ception TNO, Soesterberg.
Lee, 1.D. (2006). Driving safety. In R.S. Nickerson (Ed.). Reviews of human factors and
ergonomics. Human Factors and Ergonomics Society, St. Monica, CA.
Mckenna, F.P. (1982). The human factor in driving accidents. An overview of approaches
and problems. Ergonomics, 25, 867-877.
Mcruer, D.T., Allen, R.W., Weir, D.H. and Klein, R.H. (1977). New results in driver steering
control models. Human Factors, 19,381-397.
Michon, J .A. (1985). A critical view of driver behavior models: What do we know, what
should we do? In L.A. Evans and R.C. Schwing (Eds.). Human behavior and traffic safety
(pp. 487-525). New York: Plenum, New York.
Michon, 1.A. (Ed.) (1993). Generic intelligent driver support-A comprehensive report on
GIDS. Taylor and Francis, London.
Neisser, U. (1976). Cognition and reality: Principles and implications ofcognitive psychol-
ogy. W.H. Freeman and Company, New York.
Nilsson, L., Harms, L. and Peters, B. (2001). The effect of road transport telematics. In P-E.
Barjonet (Ed.). Traffic psychology today (1st ed., Vol. 1, pp. 265-285). Norwell: Kluwer
Academic Publishers.
Naatanen, R. and Summala, H. (1974). A model for the role of motivational factors in
driver' decision making. Accident Analysis and Prevention, 6, 243-261.
Ostlund, 1. and Peters, B. (1999). Joystick Equipped Cars' Maneuverability. VTI Medde-
lande No. 860A.
Peters, B. and Ostlund, J. (2005). Joystick controlled driving for drivers with disabil-
ities - A driving simulator experiment. Statens Vag och transportforskningsinstitut,
Linkoping,
Powers, W.T. (1998). Making sense ofbehavior - The meaning ofcontrol. Benchmark, New
Canaan, CT.
Ranney, T.A. (1994). Models of driving behaviour: A review of their evolution. Accident
Analysis and Prevention, 26, 733-750.
Ranney, T.A. (1997). Good driving skills: Implications for assessment and training. Work,
8,253-259.
Rasmussen, 1. (1986). Information processing and human-machine interaction: An ap-
proach to cognitive engineering. North-Holland, New York.
Reid, L.D. (1983). Survey of recent driving steering behaviour models suited to accident
investigations. Accident Analysis and Prevention, 15, 23-40.
Rumar, K. (1988). Collective risk but individual safety. Ergonomics, 31, 507-518.
Sanders, M.S. and Mccormick, EJ. (1993). Human factors and the automobile. In M.S.
Sanders and EJ. Mccormick (Eds.). Human Factors in engineering and design (7th ed.).
McGraw-Hill, New York.
Summala, H. (1985). Modeling driver behavior: A pessimistic prediction? In L. Evans and
R.C. Schwing (Eds.). Human behavior and traffic safety (pp. 43-65). New York: Plenum.
104 Peters, VTI and Nilsson, VTI

Summala, H. (1988). Risk control is not risk adjustment: The zero-risk theory of driver
behaviour and its implications. Ergonomics, 31, 491-506.
Taylor, D.H. (1964). Drivers' galvanic skin response and the risk of accident. Ergonomics,
7,439-451.
Vaa, T. (2005). Modelling driver behaviour on basis of emotions and feelings: Intelligent
transport systems and behavioural adaptations. In L. Macchi, C. Re and P.C. Cacciabue
(Eds.). Modelling driver behaviour in automotive environments. Office of Official Pub-
lications of the European Communities, Ispra, Italy.
Van Der Hulst, M. (1999). Adaptive control of safety margins in driving. Department of
Psychology. University of Groningen, Groningen, The Netherlands.
Weinberg, G.M. and Weinberg, D. (1979). On the design of stable systems. Wiley, New
York.
Weir, D.H. and Chao, K.C. (2005). Review of control theory models for directional and speed
control. In L. Macchi, C. Re and PC. Cacciabue (Eds.). Modelling driver behaviour in
automotive environments. Office of Official Publications of the European Communities,
Ispra, Italy.
Wickens, C.D. (1992). Engineering psychology and human performance. New York, Harper
Collins.
Wiener, N. (1954). The human use ofhuman beings. Da Capo Series in Science. Pergamon
Press, Oxford.
Wierwille, W.W. (1993). Visual and manual demands of in-car controls and displays. In
B. Peacock and W. Karwowski (Eds.). Automotive ergonomics (pp. 229-320). London:
Taylor and Francis.
Wilde, GJ.S. (1982). The theory of risk homeostasis: Implications for safety and health.
Risk Analysis, 2, 209-225.
6
From Driver Models to Modelling
the Driver: What Do We Really Need
to Know About the Driver?
OLIVER CARSTEN

6.1 Introduction
The variety of models of the driving task is almost as numerous as the num-
ber of authors who have contributed the models. Part of this variety is due
to the different applications for which the models are intended and another
part is due to the part of the driving task they are intended to describe. Since
driving encompasses so many tasks and subtasks at different levels, often per-
formed by the driver simultaneously, it is perhaps no surprise that it is hard
to find any consensus in the literature on how the process of driving should be
modelled.
Here the focus is on creating a structured model that can be used in real time, in
particular by a driver assistance system to monitor driver state and performance,
predict how momentary risk is changing, anticipate problem situations and in
response to adjust the behaviour of in-vehicle information systems and driver
assistance systems and also adjust feedback to the driver. The driver model would
therefore be the major component of a larger model supervising the interaction
among driver, vehicle and the traffic and road environment. The starting point
is a review of existing models to identify elements that can be used to predict
momentary risk.
The literature on models of the driving task is very extensive, going back at least
as far as Gibson and Crooks' Field of Safe Travel Model (Gibson and Crooks,
1938). Yet, in spite of the considerable effort put into producing such models, few
of them have been validated as predictive tools, apart from specific and limited
aspects of the driving task, where for example mathematical representations of
car following are used in microsimulation models of traffic. Still less is it the
case that any generalised model has been used in a driving assistance system to
guide the operation of the system as it supports the driver. Yet the potential of an
assistance system that 'understood' the driver is huge: it could give feedback to
novices, assist elderly drivers in difficult situations, inform a driver when he or she
is fatigued and adapt the operation of the vehicle to the needs of each individual
driver.

105
106 Carsten

6.2 A Typology of Models

Two broad types of driver model can be distinguished in the literature. The first
type is descriptive models. These models attempt to describe parts or the whole
of the driving task in terms of what the driver has to do. The second major type is
motivational models, which aim to describe how the driver manages risk or task
difficulty. The first type can be further subdivided into a number of categories -
there are task models, adaptive control models and production models.

6.3 Descriptive Models

These models attempt to describe either the whole of the driving task or some
element of it. A major feature of such models is that they are not predictive,
but are instead analytical. It is not possible to conclude from such models how
changes in driver motivation, capability or decision would affect the quality of the
performance of the driving task or situational risk.

6.3.1 Task Models

One major type of descriptive model of the driving process is the hierarchical task
model of Michon (1985). This model is shown in Fig. 6.1. The only conclusions
that can be drawn about performance or risk from this model is that a breakdown
of input, for example, environmental input at the control level or of feedback will
lead to a failure in the driving task.
Other more detailed task descriptions, such as the famous compilation of tasks
and sub-tasks by McKnight and Adams (1970), have similar qualities. They

Time Constant

General
Long
Plans

Controlled
Environmental secs
----+I Action Patterns
Input

Automatic
Environmental msec
----+I Action Patterns
Input

FIGURE 6.1. Michon's hierarchical structure of the road user task.

 6. From Driver Models to Modelling the Driver 107

d
External
Driver Disturbances
Remnant

Lateral
lateral Lateral Driver Steering - - - - - . ::~:I ....--.., Position, y
Driver Wheel Steering Steer
Position Position Operations
Operations Angle linkage Angle
Command on Functions
----+-.@~~ of Position
on Functions and Vehicle Heading
A ~y Error
of Heading Os Actuator 6 Angle.ljJ
YljJ
I Yy G
I
I
I
I
t

FIGURE 6.2. Compensatory model of driver steering (McRuer et al., 1977).

describe what the driver has to do in order to drive but they do not state ex-
plicitly how the quality of the driving performance is affected by the nature of
driver performing the tasks or by the nature of the driving being carried out. Risk
is only addressed implicitly in that a failure in performing a task is likely to lead
to a problem, but no guidance is afforded on what might cause such failure.

6.3.2 Adaptive Control Models

Another type of descriptive model describes the operation of the driving task in
terms of inputs, outputs and feedback. An example is the driver steering model of
McRuer et al. (1977) as shown in Fig. 6.2, where steering inputs are described as
compensation for errors in lateral position and heading angle. It can be noted that
this model does not account for non-deliberate errors in lateral position. Similar
models have been created for manoeuvring tasks. An example is shown in Fig. 6.3,
which is intended to describe the process of a driver approaching a T junction and
making a judgement about what speed to adopt in order to avoid collision with a
vehicle approaching from the right.
A more complex adaptive control model has recently been proposed by Holl-
nagel et al. (2003). This model covers both manoeuvring and control, as shown
in Fig. 6.4. Interestingly and presumably intentionally, this model covers only the
driver in control and not the driver who is losing or who has lost control.

6.3.3 Production Models

Tasks can also be described in terms of a formal set of rules, that is, as a production
system. Michon (1985) produced a formal set of such rules for changing gear in a
108 Carsten

Maintain
Speed

Brake
(Stop)

FIGURE 6.3. Driver decision making on approach to an intersection (Kidd and Laughery,
1964).

Anticipatory
control

FIGURE 6.4. Driver in control model (Hollnagel et aI., 2003).

 6. From DriverModels to Modelling the Driver 109

FIGURE 6.5. Action goal structure

@) for changinggear (Michon, 1985).
G=R G=1

@

G=2

@
G=3

(J)
G=4

four-speed car. The action goals structure for changing gear is shown in Fig. 6.5.
Here Action Goal 3, for example, shows the process of shifting into neutral on
stopping.
But once again such a model, while perhaps serving a purpose as a detailed
description of the purposes for which a driver changes gear, does not help to ex-
plain why one driver may choose a higher gear than another in identical traffic
circumstances. So it does not tell us in full why a driver changes gear at a partic-
ular moment. Car manufacturers know that motivation affects gear selection and
therefore have produced cars that feature automatic gearboxes that can change
style from sedate and economical to aggressive and sporty.

6.4 Motivational Models

Motivational models attempt to describe why drivers choose one alternative over
another in terms of utilities and trade-offs. In all probability, the most famous
motivation model of the driving task is the risk homeostasis model of Wilde (1982)
as shown in Fig 6.6. It is not the purpose here to provide a detailed critique of this
model - the literature is full of such critiques. More to the point is that Wilde
introduces the notion of driver capability affecting risk: perceptual skills, decision
skills and vehicle handling skills all feature in the model.
Other models introduce different aspects of driver motivation and capacity.
Naatanen and Summala (1974) add personality, experiences, motivation and vigi-
lance (see Fig. 6.7). Fuller (1984) adds capability and driver perception of rewards
and punishment (see Fig. 6.8). A more recent model by Fuller (2004) concentrates
110 Carsten

a
Expected Utilities of
Action Alternatives

4
Perceptual Skills

e
Resulting Accident
Rate
Lagged
Feedback

FIGURE 6.6. Risk homeostasis model of Wilde (1982).

on the interaction between task demands and driver capability, arguing that speed
choice in one major mechanism for the adjustment of task demands so that they
remain within capability. This latter model does not seek, however, to develop an
explanation of the major factors that determine capability; rather they are identified
in terms of broad groupings such as 'constitutional features' , that is, biological fac-
tors and 'human factors'. Focusing in particular on human factors, Rumar (1985)
describes the driving task in terms of information processing and introduces a
number of filters (physical, perceptual and cognitive) that can introduce errors. He
also incorporates such factors as motivation, experience, attention and expectation
in his model (see Fig. 6.9).

Personality Stimulus
Experiences Situation

Subj
Risk
Monitor

FIGURE 6.7. Risk threshold model

(Naatanen and Summala, 1974).
 6. From Driver Models to Modelling the Driver 111

FIGURE 6.8. Risk avoidance model (Fuller, 1984).

Reaction
Behavior

..., I

Physical Perceptual Cognitive FIGURE 6.9. Model of driv-

Filtering Filtering Filtering
ing as information processing
(Rumar, 1985).
112 Carsten

From an overall perspective, it can be seen that the motivational models introduce
more factors that permit prediction than the descriptive models. Thus, in theory,
they should be more subject to parameterisation and verification. However, such
empirical testing of the models has not generally taken place (see, e.g. Ranney
(1994), who argues that motivational models have normally not even been fully
specified let alone tested), and thus most of them remain as constructs rather than
as entities leading to the generation of rules and mathematical relationships.

6.5 Towards a Real-Time Model of the Driver

6.5.1 What Type ofModel Is Required?

In designing a model, a basic requirement is to establish how the model is to be
used and what should be its output - presuming, that is, the objective is to create
a predictive model. A second, but related issue, is that of granularity - at what
level of detail should the model operate. Here hierarchical models describing the
major elements in the driving task, such as Michon's (1985) can be useful. On the
whole, decisions related to the strategic level can be excluded as not relevant to
ADAS nor really important in determining momentary risk, although preventing
an impaired driver from setting out on a journey or even general trip suppression
or modal shift can reduce risk (see, e.g. Rumar, 1999). At the most detailed and
automated level of operation, namely vehicle control, we are faced with virtual
impossibility of predicting from one very small time step to the next what the
driver is likely to do. Such prediction would require almost absolute knowledge
about how the driver is controlling the vehicle, which would have to be based on
interpretation of physiological responses. Such prediction is beyond the capability
oftoday's cognitive science and may well be an impossible dream: humans are not
automatons whose precise reaction to a given complex situation can be predicted
with any certainty. A misinterpretation of a driver's decision could lead to severe
problems, for example preventing a lane change when the driver in the adjacent
lane has signalled consent to the manoeuvre.
This does not mean that control-level information would not provide a useful
input to the model. Thus, driver control of the vehicle could be analysed in real time
to provide the model with a depiction of a driver's characteristics. Such information
would be useful in 'tuning' adaptive systems to driver preferences. The AIDE
project has recently investigated adaptive forward collision warning systems, with
one type of adaptation being to observed driver reaction time, so that drivers who
habitually reacted quickly got later and hence less irritating warnings.
There is a strong argument, then, for focusing mainly (but not entirely) on the ma-
noeuvring level of driving. This is where many of the major risks occur - in decision
making at junctions, interactions with pedestrians, lane changing and overtaking,
etc. Errors in performing such manoeuvres are a major factor in accident causation,
so that in Great Britain 59% of all injury road traffic accidents occur at a junction
(Department for Transport, 2005). But this still begs the question of what should
be predicted. There are strong arguments for a probabilistic approach as opposed
 6. From Driver Models to Modelling the Driver 113

to a deterministic approach. Rather than predicting precisely and reliably what a

driver will do at any moment - an endeavour almost certainly doomed to failure
because of the variability of human response both between and within individuals
- a model should attempt to predict the probability of error or failure and thus
current and future risk. For example, there may be strong indications that a driver
is about to undertake an overtaking manoeuvre that, given the road layout and
traffic situation, may be highly risky. Feedback to the driver in advance of actually
starting the manoeuvre (so that the manoeuvre can be discouraged or prevented)
may well be more useful than feedback once the manoeuvre is already under way
(in the hope that it can be safely aborted). Control-level behaviour could be used
to trigger such warnings. A driver intending to overtake is likely to position the
vehicle close to or even slightly over the centreline.

6.5.2 Grouping the Factors

Thus it can be argued that an intelligent driver assistance system does not need to
fully comprehend all aspects of how a driver performs the driving task; it merely
needs to know about those driver factors that can affect the risk of vehicle operation.
This, of course, presumes that the major goal of such a system is to reduce risk,
but that seems a fairly reasonable constraint. The task of creating a model then
becomes one of identifying those driver factors that can be used to predict safety-
related performance or behaviour and of combining those factors into a structured
'model' that can be filled with rules and parameters, that is, a model that can be
empirically tested.
For this purpose, the various factors proposed in the motivational models dis-
cussed above can be identified as follows:

From Naatanen and Summala (1974)

o Personality

o Experiences

o Motivation

o Perception

o Vigilance

o Desired action

From Fuller (1984)

o Capability

o Arousal

From Rumar (1985)

o Motivation

o Experience

o Attention

o Expectation

o Perception and cognition

From Hollnagel et al. (2003)

o Goals/targets/plans
114 Carsten

Situation assessment
o

From Fuller (2004)

o Constitutional features (this model is somewhat unique in including biological

characteristics such as physical strength and reaction time)

o Training and education

o Experience

These factors can be grouped into categories as follows:

Attitudes/ personality
o Naatanen and Summala: Personality

o Rumar: Motivation

Experience
o Naatanen and Summala: Experiences

o Rumar: Experience

o Fuller (2004): Training and education

o Fuller (2004): Experience

Driver state (impairment level)

o Naatanen and Summala: Vigilance

o Fuller: Capability

Task demand
o Fuller (1984): Arousal

o Fuller (2004): Task demands

Situation awareness
o Naatanen and Summala: Perception
o Naatanen and Summala: Vigilance

o Rumar: Attention

o Rumar: Perception and cognition

o Rumar: Expectation

o Hollnagel: Situation assessment

Intention
o Hollnagel: Goals/targets/plans

o Naatanen and Summala: Desired action

We thus end up with five major categories of driver capability, performance and
behaviour that are related to risk. They are as follows:

1. Attitudes/personality
2. Experience
3. Driver state (impairment level)
4. Task demand (workload)
5. Situation awareness

Biological capability will be omitted on the grounds that permanent incapacity

relates to only a small proportion of drivers (although there may be empirical
arguments for including some aspects of age-related incapacity).
 6. From Driver Models to Modelling the Driver 115

As regards intention, this is simply a requirement by the model to predict the

driver's desired decision at a given moment. For example, we might infer from
lane position and closing towards the lead vehicle that a driver is intending to
perform an overtaking manoeuvre. The intention at one time step will then inform
the model at the next time step. The precise granularity of the model operation
will need to be defined, but it would seem sensible to start with fairly coarse time
steps to reduce the precision required.
For each of these categories, it is possible to verify from the accident literature
that they have a major role in terms of accident risk. For example, in terms of the
first category, Stradling and Meadows (2000) have cited the role of the propensity
to commit aggressive violations as a predictor of accident risk. For experience,
the models developed by Maycock et ai. (1991) relating age and experience to
accident risk can be used as evidence. For driver state, there is the literature re-
lating alcohol impairment to accident risk, including the most recent reanalysis
of the Grand Rapids data (Hurst et al., 1994). For workload, we can look at the
literature relating demands from road layout to driver performance. For the impact
of situation awareness, we can go to the studies of the impact of mobile phone use
on driving (e.g., Bums et al., 2002).

6.5.3 A Proposed Structure

A verifiable model requires a structure so that relationships can be proposed and
tested. A proposed structure, with the various categories grouped into long term
(years, months and days), medium term (hours and minutes) and short term (sec-
onds and milliseconds), is shown in Fig. 6.10.
Creating a predictive model still begs the question of what is to be predicted.
Any model created with today's knowledge must, of necessity, be probabilistic
rather than deterministic. It is not possible to create a reliable prediction of an
individual drivers decisions - after all even a drunk drive makes some correct

Task demand
(workload)

Long-term Medium-term Short-term

FIGURE 6.10. A causal structure for the categories of the driver model.
116 Carsten

FIGURE 6.11. Relationship between categories of driver factors and risk.

decisions. In addition, given sensor technology, a model running in real time cannot
be omniscient about the environmental situation and may indeed make errors in
interpreting driver actions and capabilities. So it is sensible in using some kind
of risk factor, that is, the risk of a serious error or the risk of crashing as the
dependent variable. The risk factor could be used in real time to identify when a
driver assistance system might need to warn the driver or to intervene in order to
keep prevent performance from deteriorating drastically with consequent impact
on risk. A relationship with performance and risk is suggested in Fig. 6.11.

6.5.4 Verifying the Model

Merely proposing a model does not justify it; still less does it demonstrate that the
model is useful. Thus, Grayson (1997) has stated, 'Although it would be unchari-
table to suggest that many psychological models [of driving] are little more than
the current buzz-words enclosed in boxes and joined by lines with arrowheads, it
is nevertheless difficult to avoid feeling that the term "model" has become deval-
ued in recent times'. The model proposed here is framework and not yet a usable,
validated predictive tool. Parameters for the factors need to be developed, and
the inter-relationship of the various parameters hypothesised and tested both as
regards their form (linear, exponential or power) and as regards the strength of the
relationship. Given the complexity of the proposed model, the number of relation-
ships within it (13) and the number of potential parameters, this may seem like an
impossible task. But a major consideration is that, for testing purposes, the model
can be decomposed so that not all parts are tested and verified at once. And prior
work can be found to throw light on the relationships. Thus, for example there
is already empirical evidence on the relationship between personality and driver
state in the form of studies that have examined the relationship between sensation
seeking and alcohol impairment (McMillen et aI., 1989; Yu and Williford, 1993).
Once the model has been verified or altered in parts, it should be recomposed
and tested in full. Such a test of the whole model would, of course, be a substan-
tial task requiring large numbers of drivers to be observed over a considerable
 6. From Driver Models to Modelling the Driver 117

amount of driving. But the payoff would be the delivery of a truly intelligent
co-driver.
A major step in model development and verification is identifying candidate
parameters to be included in each of the categories. Here one can conceive of a
two-stage process in which first of all the model is developed as an offline tool and
then at a second stage an online, real-time version of the model is created to run on
board the vehicle as an intelligent co-driver, supervisor of interaction with IVIS
and ADAS systems and manager of those systems to adapt to the current driver
state and current situation.
The second stage is clearly quite ambitious: care would have to be taken to
ensure that such a model did not create safety risks by going into unanticipated
states or unstable loops. One approach to minimise this would be setting boundaries
on system flexibility. Another, more straightforward alternative for an on-board
version is the creation of a simple rule-based structure that echoes the 'full' model
but does not emulate it.
However, the potential ambition of the second stage does not negate the use-
fulness of the first stage. It could have wide application and utility as an offline
design tool for new in-vehicle systems (creating 'what if?' scenarios, i.e. a kind
of failure mode and effects analysis), as a tool for evaluating road designs (e.g. in
safety audit) and even as a component in micro simulation models. In the last role,
it could replace the current rather crude practice of representing driver behaviours
by means of sampling from a set of built-in distributions, so as to represent for
example desired speed, typical headway or reaction time.
The model can initially be populated with relationships derived from the lit-
erature. Thus, since Maycock et al. (1991) have quantified the relationship be-
tween, on the one hand, experience and age and, on the other hand, risk, it may be
feasible to introduce some rules derived from that study into the model. However,
what is really required is the confirmation of parameters, conditional relation-
ships and interactions from empirical studies designed to test the hypothesised
relationships.

6.6 Developing an Online Model

A further implication of the categories is that if they are to used by a real-time
on-board system, then appropriate sensors will be required to generate information
about the current conditions, that is, the 'ifs' of the various rules. Some of that
information, for example, about driver age and experience could come from a smart
driving licence. But most of the required information will need to be observed by
the vehicle itself from how that vehicle is being driven. Thus, driver state and
appropriate methods for capturing it in real time has been investigated in a host
of European projects. However, Karel Brookhuis, one of the prime movers in this
research, has recently observed that' a valid framework for the evaluation of driver
impairment is still lacking' (Brookhuis and de Waard, 2003). Steering movements,
118 Carsten

with an increase in amplitude and a reduction in frequency indicating impairment,

are promising but to date a sufficiently reliable algorithm is lacking.
Estimating task demand in real time in order to manage driver workload has been
the focus of such projects as COMUNICAR (Amditis et al., 2002) and CEMVO-
CAS (Bellet et al., 2002). Map-based information on road type and layout can
provide basic essential information, for example on the frequency of intersections.
The fact that a driving is manoeuvring can be interpreted from vehicle yaw. Traffic
density can be provided from radar or image processing. Secondary task demand,
as opposed to driving task demand, can be inferred from interaction with enter-
tainment systems, navigation systems and other in-vehicle devices. Usage of the
mobile phone by the driver can be identified provided that there is an interface
between vehicle and mobile phone.
As regards situation awareness, the results of the HASTE project suggest that
it may be feasible to capture situation awareness from vehicle control parameters.
HASTE has been examining the impact of distraction from the use of in-vehicle
information systems on the driving task, and has identified very different impacts
from visual distraction as opposed to cognitive distraction (Carsten, 2004). Visual
distraction leads to increased lateral deviation, whereas cognitive distraction leads
to an apparent improvement in steering performance accompanied by an increase
in gaze concentration to the road straight ahead and loss of general situation aware-
ness. Thus a combination of eye movement cameras and steering sensors may be
able to provide sufficient indication of distraction and hence of loss of situation
awareness.
Steering behaviour can also potentially be used as an indicator of driving ex-
perience. Novice drivers have a generally reactive steering behaviour, whereas
experienced drivers have a more feed-forward strategy in which steering adapts to
road layout and obstacles (Jamson, 1999).
This leads to another problem that needs to be addressed. Steering behaviour
is clearly highly diagnostic, but we cannot use the same indicator, e.g. standard
deviation of lateral position, simultaneously to identify both visual distraction
and drink-driving. There is a need to investigate steering behaviour at a very
microscopic level to create unambiguous indicators. More refined indicators such
as steering reversal rate (the number of changes in steering wheel direction per
minute with a specific minimum threshold of movement) and steering entropy
(Boer, 2001) will be needed.
Finally, there is a need to identify driver intention. Salvucci and Liu (2002) have
shown how hard it is to identify a lane change that is about to be performed from
steering patterns and/or eye movements. They conclude that there is too much
variability in the data to be able to apply such methods reliably. However, there are
simpler methods for achieving this, and one that may well be effective as a signal
of a lane change is the use of the indicators by the driver. Furthermore, it could
be argued that, as in the overtaking case discussed earlier, intention need only be
known in a general rather than a precise sense. It may be more important and more
useful to know in sufficient time to deter the actual manoeuvre that a driver is highly
 6. From Driver Models to Modelling the Driver 119

likely to overtake than to detect the overtaking manoeuvre once it has started and
is too late to prevent. Here a system forearmed may mean a driver forewarned.

6.7 Conclusions
There are many driver models that are purely descriptive as opposed to being
predictive. Even the motivational models tend to be incomplete, addressing only
some of the driver factors that can elevate risk. A new structure has been proposed
here for a model that can be both verified and, if confirmed, applied in the long run to
monitor in real time the risk associated with the behaviour and performance of the
driver and to adjust feedback to the driver and tune the response of driver assistance
systems accordingly. It can be argued that, in order to produce a well-designed
advanced driver assistance system, particularly a complex multi-functional one,
such modelling is not only feasible but maybe even necessary. Otherwise, users
may reject such systems because they are not adapted to their needs.

References
Amditis, A., Polychronopoulos, A., Belotti, F. and Montanari, R. (2002). Strategy plan
definition for the management of the information flow through an HMI unit inside a car.
In Proceedings of the e-Safety Conference. Lyon.
Bellet, T., Bruyas, M.P., Tattegrain- Veste, H., Forzy, J.F., Simoes, A., Carvalhais, 1., Lock-
wood, P., Boudy, J., Baligand, B., Damiani, S. and Opitz, M. (2002). "Real-time" analysis
of the driving situation in order to manage on-board information. In Proceedings of the
e-Safety Conference. Lyon.
Brookhuis, K.A. and De Waard, D. (2003). On the assessment of criteria for driver im-
pairment: In search of the golden yardstick for driving performance. In Proceedings of
Driving Assessment 2003, the 2nd International Driving Symposium on Human Factors
in Driver Assessment, Training and Vehicle Design. Park City, Utah.
Boer, E. (2001). Behavioral entropy as a measure of driving performance. In Proceedings
of Driving Assessment 2001, International Driving Symposium on Human Factors in
Driver Assessment, Training and Vehicle Design. Aspen, Colorado.
Bums., P.C., Parkes, A., Burton, S., Smith, R.K. and Burch, D. (2002). How dangerous is
driving with a mobile phone? Benchmarking the impairment to alcohol. TRL Report 547.
TRL, Crowthome, UK.
Carsten, O. (2004). Implications of the first set of HASTE results on driver distraction. In
Behavioural Research in Road Safety 2004: Fourteenth Seminar (pp. 100-109). Depart-
ment for Transport, London.
Department for Transport. (2005). Road casualties Great Britain 2004. The Stationery
Office, London.
Fuller, R. (1984). A conceptualization of driving behaviour as threat avoidance. Ergonomics,
27, 1139-1155.
Fuller, R. (2004). Towards a general theory of driver behaviour. Accident Analysis and
Prevention, 37,461-472.
120 Carsten

Gibson, J.1. and Crooks, L.E. (1938). A theoretical field-analysis of automobile driving.
American Journal of Psychology, 51,453-471.
Grayson, G.B. (1997). Theories and models in traffic psychology - A contrary view. In
T. Rothengatter and E. Carbonell Vaya (Eds.). Traffic and transport psychology: Theory
and application. Pergamon, Amsterdam.
Hollnagel, E., Ndbo, A. and Lau, LV. (2003). A systemic model for driver-in-control. In
Proceedings of Driving Assessment 2003, the 2nd International Driving Symposium on
Human Factors in Driver Assessment, Training and Vehicle Design (pp. 87-91). Park
City, Utah.
Hurst, P.M., Harte, D. and Frith, W.1. (1994). The Grand Rapids dip revisited. Accident
Analysis and Prevention, 26, 647-654.
Jamson, A.H. (1999). Curve negotiation in the Leeds driving simulator: The role of driver
experience. In D. Harris (Ed.). Engineering psychology and cognitive engineering (Vol.
3, pp. 351-358). Ashgate, London.
Kidd, E.A. and Laughery, K.R. (1964). A computer model of driving behaviour: The
Highway Intersection Situation. Report VI-1843-V-I, Cornell Aeronautical Laboratories,
Buffalo, New York.
Maycock, G., Lockwood, C.R. and Lester, IF. (1991). The accident liability of car drivers.
Research Report 315. Transport and Road Research Laboratory, Crowthorne, UK.
Michon, lA. (1985). A critical review of driver behaviour models. In L. Evans and R.G.
Schwing (Eds.). Human behavior and traffic safety (pp. 485-520). Plenum Press, New
York.
McKnight, AJ. and Adams, B.B. (1970). Driver education task analysis. Vol. I: Task de-
scriptions. Human Resources Research Organization, Alexandria, Virginia. Final Report,
Contract No. FH 11-7336.
McMillen, D.L., Smith, S.M. and Wells-Parker, E. (1989). Brief report: The effects of
alcohol, expectancy, and sensation on risk-taking. Addictive Behaviors, 14, 477-483.
McRuer, D.T., Allen, R.W., Weir, D.R. and Klein, R.R. (1977). New results in driver steering
control models. Human Factors, 19,381-397.
Naatanen, R. and Summala, H. (1974). A model for the role of motivational factors in
drivers' decision-making. Accident Analysis and Prevention, 6, 243-261.
Ranney, T.A. (1994). Models of driving behaviour: A review of their evolution. Accident
Analysis and Prevention, 26, 733-750.
Rumar, K. (1985). The role of perceptual and cognitive filters in observed behavior. In
L. Evans and R.G. Schwing (Eds.). Human behavior and traffic safety (pp. 151-165).
Plenum Press, New York.
Rumar, K. (1999). Transport safety visions, strategies and targets: Beyond 2000. European
Transport Safety Council, Brussels.
Salvucci, D.D. and Liu, A. (2002). The time course of a lane change: Driver control and
eye-movement behaviour. Transportation Research Part F, 5, 123-132.
Stradling, S.G. and Meadows, M.L. (2000). Highway code and aggressive violations in UK
drivers. In Global Web Conference on Aggressive Driving. http://www.aggressive.drivers.
com/papers/stradling-meadows/stradling-meadows. pdf.
Wilde, G.1.S (1982). The theory of risk homeostasis: Implications for safety and health.
Risk Analysis, 2, 209-225.
Yu,1. and Williford, W.R. (1993). Alcohol and risk/sensation seeking: Specifying a causal
model on high-risk driving. Journal ofAddictive Diseases, 12(1),79-96.
 III
Learning and Behavioural Adaptation
7
Subject Testing for Evaluation of Driver
Information Systems and Driver
Assistance Systems - Learning Effects
and Methodological Solutions
KLAUS BENGLER

SUMMARY
Nowadays, subject testing represents a well-established methodology to evaluate
different properties of driver information systems and driver assistance systems.
Among several criteria, learnability is one important system property. User and
usage strategies are dependent on the subject's learning state, for example, to
switch attendance between driving task and operation of a driver information
system. Therefore it is wishful that the user acquires a model of the system, for
example learns as quickly as possible. Also, the intended usage of driver assistance
systems in given driving situations is influenced by the user's experience. A suitable
way to investigate related questions is to conduct a typical learning experiment
and to analyse data with the given methodology. In this type of experiment, the
familiarity and training state of the subject are set as independent variable. Beside
learnablity, other properties of human-machine interaction are to be investigated
and evaluated. In this case, however, the learning effectuated by a subject is an
important dependent variable or even noise in sense of measurement theory that
might cover a given main effect. After some empirical examples, possible solutions
will be discussed that help to manage this problem with justifiable expense.

7.1 Introduction
Following the idea of user-centered system design, subject testing is an estab-
lished and frequently used procedure to evaluate and ensure quality of human-
machine interface (HMI) concepts. Appropriate methods are applied in areas of
In-vehicle information systems (IVIS)l and advanced driver assistance systems
(ADAS 2 ) (Mayser, 2002). A good overview of different methodologies and of

1 Also called driver information systems (DIS).

2 Also called driver assistance systems (DAS).

123
124 BENGLER

their interaction with possible product development processes is given in the pro-
ceedings of Bundesanstalt fur StraBenwesen (2000).
Documents like DIN 66234 Part 8 (DIN66234 1986) and the European Statement
of Principles (2000/53/EC 2000) make clear that beneath properties like 'error
robustness', 'interruptability' or 'visual demand', learnability of a system is also
an important system property. More and more questions are raised concerning the
quality of subject testing in the sense of test theory and methodology (Kanis, 2000;
Haigney 2001). This paper discusses the conditions under which an independent
evaluation of dialogue quality and assistance characteristics can be conducted. An
evaluation that accounts for the fact that learnabilty and learning state should not
influence the results erroneously.
Frank and Reichelt (2001) also mention that learnability is one criterion that
can be tested during the development of an ADAS by expert judgement and ex-
perimental testing to ensure high product quality and system acceptance.
In general, learning is defined as a 'permanent change of behaviour based on
experience' (Hilgard and Bower, 1966) and therefore parameters for the learning
process were modelled for given tasks. More specifically, Woods (1999) states that
the user is building up heuristics and simplified models to structure the interaction
with a given system.
In addition, Reeves (1999) introduces the concept of 'cognitive complexity'
as quality for an HMI based on models for cognitive processes. Cognitive com-
plexity describes the property of a system that enhances or prohibits the user's
learning process. This process includes elements of perception, model building
and categorisation. Following this, Reeves (1999) gives recommendations for sys-
tem interaction concepts and information presentation to increase the usability
of an HMI concept. These include guidelines for a leamer-oriented design pro-
cess that targets interaction concepts that support perception, visualisation, model
building, categorisation and problem solving (cf. also Groeger, 1991).
This emphasises that learnability of a system is a predominant feature of an
interaction concept besides other properties that contribute to usability and utility
of the system. Therefore, in the following, learning experiments as one type of
experiment will be distinguished from system evaluations.
A learning experiment typically focuses on questions about acquisition of
knowledge on the system and users' mental models. Questionnaires and inter-
views are used as dependent variables as well as performance measures (e.g.
driving performance, user errors). In case of a system evaluation, properties like
visual or motoric demand are analysed as well as additional performance measures
and user errors - that is observable behaviour (Woods 1999) describes the idea
of so-called built in or designed system diseases as one of the most important
source for human errors in contrast to the mostly stated human error and give more
information on error analysis as a further evaluation method. Thus, in both types
of experiments, the analysis is based on similar measurements. If a subject was
learning how to operate the system during a system evaluation, the learning pro-
cess will produce variability and cover the main effect and item of interest of this
investigation.
 7. Subject Testing for Evaluation of Driver Information Systems 125

In this context, subjects' learning capabilities and inter-individual differences

playa very important part of the learnability of the system. One must not oversee
that in most cases experiments and data gathering take place in an automotive
environment (vehicle or driving simulator) that is not familiar for the subject.
Some examples shall help to get a better understanding of the quality and range
of learning effects after some methodological discussions.

7.2 Methodological Issues

In general the in-car usage situation that is the learning context for a given in-car
HMI is described with the following model: The driver has to fulfil the primary
driving task that can be subdivided in stabilisation, manoeuvring and navigation.
While stabilisation includes highly automated activities on the skill level, manoeu-
vring and navigation afford more rule- and knowledge-based user behaviour. Also
these three subtasks and the acquisition of skills, rules and knowledge are sub-
ject to learning experiments. But the focus of the given publication is secondary
and tertiary tasks include the operation of driver information and driver assistance
systems like audio, navigation, HVAC and ACC.
Obviously primary, secondary and tertiary tasks use the same limited motoric,
sensory and cognitive resources of the driver. The design of in-car HMI therefore
requires that the primary task does not crucially interfere with a secondary or
tertiary task. Apparently all these tasks underly learning processes that contribute
to increasing driver performance. In the case of the primary task this is pointed
out by a decrease in accident risk with an increase of driving experience. The
content learnt by the driver is very complex as it includes procedural knowledge,
semantic knowledge as well as knowledge on system behaviour and system limits.
The systems that construct these learning 'items' range from the traffic system in
general over traffic situations, the vehicle in its static and dynamic properties up
to devices and functionalities.
To acquire this knowledge, drivers apply strategies that are well known from
other domains: especially, rules on the traffic system and basic interaction with
the vehicle. Learning strategies are learnt by education and observation and after
this refined by training/exercise. Operation of in-car functionalities is, to a high
degree, based on the application of transfer of trial-and-error strategies. Manuals
might also playa minor role.
Manstetten (2005) emphasises the concept of learnability of driver assistance
systems:

With reference to the project RESPONSE, 'a system is learnable, if accurate

assimilation of information by the driver occurs, evidenced in the driver's un-
derstanding of system function, system handling and situational limits ' .
A 'self-explanatory' support system is defined as a 'driver assistance system
leaving a minimal amount of learning demand to the driver and eliminating
learnability issues which can result in safety-critical traffic situations' .
126 BENGLER

This makes clear that learning processes expand the range of the driver especially
by the establishment of highly automatised expert behaviour. Vehicle stabilisation
(distance and lane keeping) and the more efficient usage of in-car devices based on
usage strategies are an example. An example for such a strategy is to use a given
device not in any but a suitable traffic situation or to interrupt the operation and
continue afterwards (Sayer et al., 2005).
Learnability is therefore a beneficial system property that is to be tested
('Learnability' DIN 66234/8) like other ergonomic qualities during product de-
velopment. A problematic fact in this context is that only few models ex-
ist' that describe the learning behaviour described in the section above. This
fact makes it difficult to plan and conduct learning experiments during system
development.
On the other hand, learning is also a potential source of 'noise' in empiri-
cal testing to other values describing HMI qualities like visual demand or in-
terruptability. Therefore, learning processes have to be taken into account ei-
ther as a main effect or as a side effect at the different steps of the evaluation
process:

planning (sample construction, subject selection, procedure, scenarios);

conduction (training procedure, training criterion, measurements);
analysis (data qualification, interpretation, explanation).

This is now the point to stress the lack of detailed learning models that could
help the experimenter to decide on the above questions.
Especially the model of Rasmussen (1983) is suited and used to describe learning
in the context of the driving task.
In relation to the driving task, learning is described as a staged process
Rasmussen (1983), however, the transition between skill-based - rule-based -
knowledge-based stages are not described very concisely. Therefore, the stages
are difficult to handle for planning and experimental procedures. But they can be
used post hoc to reduce variability, in most cases, as a covariate.
The learning processes in relation to in-car HMI are mostly modelled using
the power law. This shall also help to deal with these effects within experimental
procedures. As different system types and usage scenarios might require a very
differentiated discussion, mostly the necessary parameters to describe the power
law function are missing.

7.3 Experimental Examples

The following section will give examples for learning experiments and show that
there is a possibility to describe in-car learning behaviour. Established methods can
be used for this purpose. The description and modelling of these specific learning
processes could then help to plan and conduct future evaluations of comparable
functionalities.
 7. Subject Testing for Evaluation of Driver Information Systems 127

7.3.1 Evaluation ofa Multimodal HMI

Strategies to switch between secondary task and driving task are highly dependent
on the learning state. Therefore, learning processes are an important factor during
the evaluation of driver information systems. This can be seen looking at the
evaluation of multimodal HMI demonstrator and error analyses on interaction
data (cf. Bengler, 2002).
Within a field test, learning of a multimodal HMI that allowed to operate a
navigation system and an in-car phone was investigated. Subjects could change
anytime during operation between input modalities (speech vs manual turn/push
knob). The study investigates user behaviour as well as driving behaviour in dif-
ferent traffic situations. Specific interest is put on learning process and effects that
are correlated with learning.
The questions were as follows:
Is the user able to operate typical driver information systems multimodally while
driving?
Is it possible to acquire necessary knowledge in appropriate time (i.e. to learn
an efficient way of multimodal interaction)?
The study was conducted as a 2 x 2 within-subject design with independent
variables secondary task (telephone/navigation) and primary driving task (sim-
ple/complex). The telephone task included to dial an instructed phone number
while driving. The navigation task required to enter a destination (city and street)
while driving. Subjects had free choice of the input modality (manual, voice). The
simple driving condition was driving on a straight one or two/lane road, with right
of way and only few traffic lights (the complex section was on narrow curvy urban
roads without lane markers including yield/pass situations).
To negotiate the complete test track (12 km) required about 25 min. All subjects
(N = 11) had a valid driving license, continuous driving experience and no or only
minimal experience with speech recognition systems.
After a very short introduction and a very short training using the speech recog-
nition system, each subject had to go through the test course three times: once
without and twice with secondary task operations, following the instructions of
the experimenter.
Data during voice dialling at the beginning and at the end of the experiment
were used to analyse learning effects.
Driving performance and in-car operations of the subjects were videotaped.
Considering the first question, all subjects safely negotiated the secondary tasks,
and compared to other studies, speed was not significantly reduced during sec-
ondary task operation.
The videotaped data were analysed following an error taxonomy that differen-
tiates the following:
Timing errors: (user speaks to early):
User speaks to non-active ASR
User speaks while system's speech output
128 BENGLER

40

35

30

25

20

15

10

o-J,.-"
Timing Vocabulary Dialog Recognition
Errors Errors Errors Errors

FiGURE 7.1. Frequency of different error types during first and last phases of experiment
(all subjects and traffic situations).

Vocabulary errors : Non-valid/available command word used; Irrelevant item

used (e.g. thinking aloud, laughing, hesitations 'aahm'); Inadequate command
(e.g. ' DELETE' deletes all digits, instead 'ERROR' deletes only last digit) .
Dialog errors : User ignores system prompt; User interprets system prompt in
the wrong way.
Recognition errors: ASR recognises a correct user input in the wrong way; ASR
'ignores' a correct user input.
The error analysis shows that there is a well-known but remarkably fast learning
effect that can be seen in the subjects ' error behaviour (cf. Fig. 7.1) .
A significant error reduction in all error categories can be seen . This effect is
not based on a changed mode of operation of the speech system under test, but
rather on a changed usage and speech behavior of subjects . This means that appro-
priate command words are spoken at the right point of time . Within the dialogue,
questions from the system are answered correctly by subjects . The reduction of
recognition errors refers to the fact that subjects corrected their level of speech
volume . It can be shown that on the one hand the number of successfully finished
dialogues increases due to a change of behavior, and on the other hand, a very
strong learning effect using speech-operated systems can be observed. In case of a
system evaluation using a treatment; - treatments design - comparing a speech-
operated HMI A to a second speech-operated HMI B - the second treatment under
test would profit the subjects' experience with the first treatment. Assuming that
both speech-operated systems would 'behave' the same way regarding vocabulary,
timing and dialogue.
Analysis of total task times presents a similar result. During the experiment, total
task times are remarkably reduced (Table 7.1). A more detailed analysis shows that
a reliable estimation of total task time is possible after the second run.
 7. Subject Testingfor Evaluation of DriverInformation Systems 129

TABLE 7.1. Totaltask times (mean and standarddeviation) telephone and navigationtasks
in four subsequentruns during the experiment (N == 11).
Telephone tasks Navigation tasks

2 3 4 2 3 4

Mean 104.4 40.7 38.9 41.0 56.7 35.6 40.2 36.0

SD 98.06 21.48 18.31 21.00 34.13 7.45 16.23 14.46

7.3.2 Destination Entry While Driving

A second experiment shall outline that learning effects have to be taken into ac-
count if dialogue properties like workload or demand that are often based on total
task time, total glance time or single glance time are evaluated. For these vari-
ables, precise criteria values are formulated for evaluation. The critical question
is whether only the 'worst case' of an novice user should be tested or also the
efficiency potential of a given dialogue performed by a trained user. Because the
goal should be that a dialogue is performed in the beginning, but performance
increase is given with increasing user experience and practice.
Destination entry shall serve as a secondary in-car task and an example for
a dialogue of long total task time. Within a learning experiment, Jahn (2002)
compared total task times of destination entry dialogues, using different navigation
systems. Of interest is whether a learning process takes place and which parameters
describe the learning process of this specific dialogue in the best way.
Driving on a test track, subjects were instructed to enter a destination consisting
of city and street. The system was operated using a turn push button.

Twelve subjects participated in the experiment (aging 35-55 years) having more
than 100,000-km driving experience.
In sum, 100 destination entries were negotiated at speeds of 40-50 km/h.
Two route complexities are distinguished.
Easy: 1.5 km, straight
Winding: 1.2 km; turning into narrow roads
The resulting data set gives insight into the learning process of this very specific
task.
At first, a learning process can be seen with a clear performance improvement
between Blocks 1 and 2 for the winding route. In addition, there is a significant
influence of route complexity on mean duration for destination entry and the level
of the learning process in general (see Fig. 7.2).
Comparison of different subjects shows two further effects. Learning processes
of subjects 1 to 5 start at significant different levels and are hardly comparable in
their shape. This means that either the mean value computed of this sample would
go along with a high standard deviation. This is especially problematic if the data
would be used for comparison with another sample using a different system. Or,
the experimenter would have to decide to extract subgroups of subjects due to their
difference in learning behaviour.
130 BENGLER

130 r---T-----r----r----r--~-___,____, FIGURE 7.2. Mean duration of a destination

120 entry depending on the experimental session
~ 110 and route characteristic. Block 1 = 20 desti-
~
c: 100 nation entries. Block 2-6 = at times 16 desti-
Q) 90 nation entries.
80
70
... -....-_.---- --.
60
50
40
30
20 winding - -
easy ---------
10
o '-----"_----'-_--'--_-'----------.L._---'-----'

2 345 6
Block

Figures 7.2 and 7.3 show that both effects -learning process and inter-individual
differences - result in a high level of variance of the data set. Therefore, they should
be treated as error variance in system evaluation experiment.

7.3.3 Evaluation ofDriver Assistance Systems

In the area of empirical evaluation of driver assistance systems, learnability and
learning effects are an important parameter. Simon and Kopf (2001) emphasise this
vividly in the context of automatic cruise control (ACC) systems. The study con-
ducted was a long-term field trial to investigate learning effects. Five subjects who
participated used an experimental car equipped with ACC for 3 weeks for their ev-
eryday trips. Data show that system experience increases with usage time. But also
that subjects dealing with the limits of the ACC system establish different strategies.

130
120
~ 110
e-
C 100
<D
90
en<D 80
"0
70
'0
c 60
0
~ 50
:; P1
"0
40 P2 ---------.
C 30 P3 -----------
ctj
<D 20 P4 FIGURE 7.3. Mean duration of a destination
~ P5 _._._._._--
10 entry depending on experimental run per
0 subject (PI to P5). Block I = 20 destination
2 3 4 5 6 entries. Block 2-6 = at times 16 destination
Block entries.
 7. Subject Testing for Evaluation of Driver Information Systems 131

The authors distinguish 'eager testers' and 'careful approachers'. Obviously, dif-
ferent usage strategies again lead to a remarkable variance within this small sample.
This variance can be used to describe different learning strategies. But driving per-
formance and other effects - especially at the beginning of the experiment - have
to be interpreted by taking account of these individual learning strategies.
An attempt to present a model on skill acquisition of an ACC system is given by
Hoedemaeker (1999). She proposes the idea that the adaptation of the ACC mental
model results of two processes that operate at different abstraction levels within
the mental model. The first process is based on the difference of expected system
behaviour provided by the user's mental model of the ACC and the observed system
behaviour. Depending on the degree of wrong predictions, the ACC mental model
is updated to arrive at a more accurate account of the ACC's operational domain.
The second process dealing with the adaptation of the mental model takes place
at a higher abstraction level. The intensity of ACC usage then is highly correlated
with the degree of perceived inconsistencies between ACC behaviour and user
expectation based on his or her mental model. The less inconsistencies the more
system usage and vice versa.
A very sophisticated long-term field operational test using an ADAS was con-
ducted by Weinberger (2001). The experiment gathered more information about
the learning process for the usage of controls and display and the judgement of
take-over situations. Participants used an ACC-equipped car for a period of 4
weeks per person. Data were analysed with respect to the duration of the learning
phase. The change in behaviour was investigated using drivers' self-assessment of
the length of learning and observed driving behaviour during take-over situations.
The results suggest that 2 or 3 weeks are needed to learn the operation of ACC and
the assessment of take-over situations for a goal-directed system usage. Interesting
is the methodological advice given by the authors that other ACC users might need
a different learning time as the participants in the study drove much more than the
average driver.
Kostka et al. (2004) recommend the analysis of driver errors by expert observa-
tion as a practical empirical method to evaluate workload and distraction but might
serve as well as a tool to investigate the learnability of the concept and especially
erroneous user expectations on system functionality.
Results from this study and that of Weisse (2002) give the impression that the
investigation of learning processes and individual behaviour in the context of driver
assistance systems demands considerably more effort than driver information sys-
tems.

7.4 Solutions
The discussion of the 'learning problem' and the selected experiments show that on
the one hand learning processes and learnability in the in-car domain can be inves-
tigated using classical methodologies. Due to this method, the effort is remarkable.
132 BENGLER

On the other hand it is also true that learning processes can lead to considerable
data variablity in evaluation experiments:

learning and carry -over effects;

positive or negative transfer in within-subjects designs treatment A, which pre-
cedes treatment B in the experiment, will have influence on the variability of
values in the second measurement; and
everyday knowledge and experience will interact with subjects' performance.

Unfortunately, it is not possible to compute or estimate error variance as expe-

rience and concise models for these effects are still missing. This procedure might
become feasible if the characteristic learning curve was known for a given system.
After the citations of learning experiments and related problems, some approaches
shall be presented to control learning effects in system evaluations.
Inter-individual differences should be covered most efficiently using a within-
subjects design. A systematic distribution of subjects on experimental treatments
using a Latin Square can avoid or at least balance positive and negative transfer
effects over the whole experimental data set. Using this approach, the resulting
number of subjects must not be underestimated (Bortz, 1999).
The most reliable solution of the learning problem is to use highly trained
subjects combined with the application of a standardised performance criterion
before measurement begins. This approach, however, requires a substantial effort
for training and testing. The study of Jahn et al. (2002) gives an impression of the
necessary training effort for a destination entry example.
A further possibility is to use so-called 'expert users' as subjects, for example,
for conduction of a destination entry study, a sample of experienced users who
have everyday experience with the given system, however, in this case also, a
performance test should be conducted. But especially for the investigation of novel
usage concepts this approach is no alternative.
A general recommendation on the conduction of learning experiments with
driver information systems is given by Rauch et al. (2004). They recommend that
learning experiments that investigate in-car systems for use while driving should
be conducted in a dual-task setting to increase the validity of the method.

7.5 Conclusions
The examples give an impression of learning effects that have to be expected
in connection with subject tests evaluating driver information systems and driver
assistance systems. It turns out that learning experiments that investigate technolo-
gies such as voice recognition and future driver assistance systems would ease the
planning and conduction of system evaluation and raise their quality.
In addition, one can state that the investigation of total task time and other
performance measures is only reliable and makes sense if learning is finished and
subjects are in a stable state.
 7. Subject Testing for Evaluation of Driver Information Systems 133

Suitable models are still missing that would describe learning of driver informa-
tion systems and driver assistance systems in detail. Therefore, the experimenter
has to reduce high variances that are based on learning effects by carefully selecting
experimental procedures that do not exceed manageable effort.

References
2000/53/EC (2000). 2000/53/EC: Commission recommendation of 21 December 1999 on
safe and efficient in-vehicle information and communication systems: A European state-
ment of principles on human machine interface. Official Journal of the European Com-
munities L19 43 (25.01.2000), 64-68.
Bengler, K., Noszko, Th. and Neuss, R. (2002). Usability of multimodal human-machine-
interaction while driving. In Proceedings of the 9-th World Congress on ITS, CD-Rom,
9th World Congress on Intelligent Transport Systems, ITS America, October 2002.
Bortz, J. (1999). Statistik fur Sozialwissenschaftler (5th ed.). Springer-Verlag, Berlin-
Heidelberg.
Bundesanstalt fur StraBenwesen (2000). Informations- und Assistenzsysteme im Auto be-
nutzergerecht gestalten. Methoden fur den EntwicklungsprozeB. Bd. Heft 152. Bergisch
Gladbach: Bundesanstalt fur StraBenwesen, 2000.
Norm DIN66234 8 8 (1986). Bildschirmarbeitsplatze: Grundsatze der Dialoggestaltung.
Frank, P. and Reichelt, W. (2001). Fahrerassistenzsysteme im Entwicklungsprozess. In Th.
Jurgensohn and K.-P. Timpe (Hrsg.) (2001). Kraftfahrzeugfuhrung. Springer, Berlin,
Heidelberg, New York, pp. 71-80.
Groeger, J.A. (1991). Learning from learning: Principles for supporting drivers. In Inter-
national Symposium on Automotive Technology & Automation (24th: 1991: Florence,
Italy). Croydon, UK: Automotive Automation, pp. 703-709.
Haigney, D. and Westerman, SJ. (2001). Mobile (cellular) phone use and driving: A critical
review of research methodology. Ergonomics, 44(2),132-143.
Hilgard, E.R. and Bower, G.R. (1966). Theories of learning (3rd ed.). Appleton-Century-
Crofts, New York.
Hoedemaeker, M. (1999). Driving Behaviour with Adaptative Cruise Control and the Ac-
ceptance by Individual Drivers. Delft University Press, Delf.
Jahn, G., Krems, J.F. and Gelau, C. (2002). Skill-development when interacting with in-
vehicle information systems: A training study on the learnability of different MMI con-
cepts. In D. de Waard, K.A. Brookhuis, J. Moraal and A. Toffetti (Eds.). Human fac-
tors in transportation, communication, health, and the workplace. Shaker Publishing,
Maastricht, The Netherlands.
Kanis, H. (2000). Questioning validity in the area of ergonomics/human factors. Er-
gonomics, 43(12), 1947-1965.
Kostka, M., Dahmen-Zimmer, K., Scheufler, I., Piechulla, W. and Zimmer, A. (2004).
Fahrfehlerbeobachtung durch Experten. Eine Methode zur Evaluierung von Belastung
und Ablenkung des Fahrers. In A.C. Zimmer, K. Lange, et al. (Hrsg.). Experimentelle
Psychologie im Spannungsfeld von Grundlagenforschung und Anwendung. Proceedings
43. Tagung experimentell arbeitender Psychologen (CD-ROM). Regensburg.
Manstetten (2005). Evaluating the traffic safety effects of driver assistance systems. AAET
2005, Automation, Assistance and Embedded Real Time Platforms for Transportation:
Airplanes, Vehicles, Trains, 6th Braunschweig Conference. (Vol. 1). Braunschweig, DE.
134 BENGLER

Mayser, C. (2002). An advanced concept for integrated driver assistence systems

(S.A.N.T.O.S.-Projekt). Proceedings of the 9-th World Congress on ITS, CD-Rom, 9th
World Congress on Intelligent Transport Systems. ITS America, October 2002.
Rasmussen, J. (1983). Skills, rules and knowledge; signals, signs and symbols and other
districtions in human performance models. IEEE Transactions on Systems, Man and
Cybernetics, SMC 13(3).
Rauch, N., Totzke, I. and Kruger, H.-P. (2004). Kompetenzerwerb fur Fahrerinformation-
ssysteme: Bedeutung von Bedienkontext und Menustruktur, In VDI-Berichte Nr. 1864.
Integrierte Sciherheit und Fahrerassistenzsysteme. VDI-Verlag, DUsseldorf.
Reeves, W. (1999). Learner centered design. Sage Publications,Thousand Oaks, CA.
Sayer, J.R., Devonshire, J.M. and Flannagan, C.A. (2005). The effects of secondary tasks
on naturalistic driving performance. Report No. UMTRI-2005-29. The University of
Michigan TransportationResearch Institute, Ann Arbor, Michigan, 48109-2150.
Simon, J. and Kopf, M. (2001). A concept for a learn-adaptive advanced driver assis-
tance system. In CSAPC 2001, 8th Conference on Cognitive Science Approaches to
Process Control, The Cognitive Work Process: Automation and Interaction, Universitat
der Bundeswehr, Neubiberg, Germany, 2001, pp. I-II.
Weinberger, M. (2001). Der EinflufJ von Adaptive Cruise Control Systemen auf das
Fahrerverhalten. Shaker Verlag, Aachen.
Weisse, J., Landau, K., Mayser, C and Konig, W. (2002). A user-adaptive assistance sys-
tem. In Proceedings, ORP2002 - 2nd International Conference on Occupational Risk
Prevention, February 2002.
Woods, D.D. and Cook, R.I. (1999). Perspectives on human error. Hindsight biases and
local rationality. In F.T. Durso (Ed.). Handbook of applied cognition. John Wiley and
Sons, Riverton, NJ.
8
Modelling Driver's Risk
Taking Behaviour
WIEL JANSSEN

8.1 Introduction
A realistic estimate of the risk-reducing effects that will be gained by the intro-
duction of advanced driver supports (ADAS) requires knowledge of a number of
elements, of which user behaviour is the least understood. This paper focuses on
some of the most essential knowledge that is already available, in particular on the
mechanisms by which users could possibly change their behaviour once they start
using the support.

8.2 Expected Risk Reductions from New Technology

on the Road
The safety effects resulting from the introduction of new technology in vehicles
or - for that matter - in the road infrastructure follow from the interaction of four
components:

1. The so-called 'engineering estimate' or intrinsic effectiveness estimate of a

device's expected safety effect, that is, the accident reduction to be expected on
the basis of purely statistical or mechanical considerations. For example, the
seat belt's effectivity in increasing the probability of surviving a vehicle crash
is estimated to be around 40% to 50%. This would then be the initial estimate
of the reduction in fatalities if the entire population would use the belt.
2. The degree of penetration or use rate of the device in the relevant population. For
devices which rely, for their effectiveness, on the acceptance by the population,
there is the issue of selective recruitment, which means that the use rate per se
and/or the effect a measure achieves is affected by self-selective processes in
the population. The hypothesis is that those who opt for the device differ from
those who do not in respects that are essential to measure the effectiveness, the
particular assumption being that those who are least inclined to accept a safety
device would profit the most from it (e.g. Evans, 1985).

135
136 Janssen

3. Effects on and changes in the user's behaviour that may be brought about by
the device, in particular the so-called behavioural adaptation processes.
4. The functional relationships linking behavioural parameters (e.g. driving speed
and its variability) to resultant accident probability and severity.

We need to know more on all of these, but this paper focuses on the last two
and on behavioural adaptation in particular. From the behavioural point of view,
selective recruitment is an almost equally important issue, but it deserves more
space than could be devoted here.

8.3 Behaviour When Driving with Supports

A number of behavioural factors are critical to the success of a support system

(Janssen, 2001).

8.3.1 The Importance of Plain Old Ergonomics

The first factor is the badly designed in-vehicle (or roadside) supports, which
may cause mental over- or under load of drivers, both of which may lead to
a decrease in situational awareness and to increased risks. Even recent, on-the-
market, systems sometimes suffer from this (Janssen et al., 1999). This should
no longer be acceptable, given the extensive knowledge on display and han-
dling characteristics, colour use, lettering, etc., that is contained in ergonomic
handbooks.

8.3.2 The Loss of Potentially Useful Skills

When drivers are relieved from executing certain elements of the driving task,
they may lose the ability to perform the associated skills. This is the equivalent of
certain everyday phenomena, like younger people in Western societies supposedly
no longer being able to do even simple additions without the help of a calculator.
That skills are lost may not matter at all in some cases and it is certainly not
necessary to retain each and every antique skill for some doomsday to come;
however, sometimes the skill that is gone could well have been a safety asset in
some situations.

8.3.3 Opportunities for New Errors

While advanced supports may take the opportunity away for drivers to make er-
rors, the design and the required maintenance of these systems may themselves
constitute new sources of errors. This also needs to be taken into account when
estimating net safety effects in, at least, a qualitative sense.
 8. Modelling Driver's Risk Taking Behaviour 137

8.3.4 Problematic Transitions

It is generally recognised that the introduction of (partially) automated supports
will require a solution for problems associated with the ensuing mix of vehicles
that have the support and those that have not. This is the problem of a transition
taking place in time.
Less attention has been directed to the problems a driver may experience when he
or she has to make a transition in space, that is, from non-automated to automated
parts of a road network and vice versa. In particular, adaptation and take-over
effects from the automated environment may play a negative role when getting
adjusted to the non-supported environment again.
On a more general level, the driver may actually have to learn dealing with
advanced supports. This may require fundamental changes in training curricula,
as they presently exist.

8.3.5 Risk and Risk Perception: My Risk and Yours

People accept less risk that is imposed on them by others than when they make
the choice themselves. Slightly exaggerating, mountaineers may be more afraid
during the flight to the Himalayas than when climbing Everest itself. Thus, ex-
tra safety has to be provided whenever drivers feel an automatic device takes
the process of determining what is risky out of their hands. This is a some-
what self-defeating process from the point of view of introducing an automated
environment.

8.4 Behavioural Adaptation

'Behavioural adaptation' is a summary descriptive term that stands for a number
of phenomena that may occur as a consequence of drivers interacting with the
newly introduced element in their task environment. The general connotation of
the concept is that it is detrimental to the positive effects originally foreseen to
result from the new support system. Two forms may be distinguished, direct and
higher-order behavioural adaptation.

8.4.1 Direct Changes in Behaviour

It has been established that drivers do indeed show riskier behaviours in several
important cases in which risk-attenuating devices were provided. One of these is
ABS, which was studied in the famous Munich taxi driver experiment (Aschen-
brenner et al., 1994). This study had both a retrospective and a prospective part,
and in neither could it be shown that ABS-vehicles were involved in less accidents
than standard vehicles. The other case is the one involving seat belts (Janssen,
1994a). Table 8.1 shows the results of an instrumented-vehicle study with subjects
138 Janssen

TABLE 8.1. Increase in average driving speed on motorway and in amount of

car-following at very short headways, over three consecutive measurements at about
4-month interval in first year of belted driving, compared to previously unbelted driving
(baseline) .
1st 2nd 3rd

Increase in driving speed (km/h; baseline: 112.3 kmlh) +2.2 +2.6 + 2.8
Increase in occurrence of following headways < 0.5 s +0.5 + 6.3 +7.9
(%; baseline: 5.5 %)

who originally were non-wearers of the belt and who became wearers for the pur-
pose of the study. This long-term study could apply an almost ideal experimental
design in that it had an alternating own vehicle/instrumented vehicle design; that
is, subjects started wearing the belt all the time in their own vehicle and came to
the laboratory for measurements in an instrumented vehicle at regular (4-months)
intervals after an initial beltless baseline measurement. Their driving performance
was compared to a control group of non-wearers who remained so for at least the
duration of the study.
While ABS and seat belts are apparently demonstrations of the existence of
the phenomenon, it is not yet clear (a) whether this will always happens or what
would distinguish cases in which they do from cases in which they do not and (b)
whether the compensation is complete and will totally eliminate the safety effect
that should follow from the engineering estimate.
To come to terms with these questions, we would need valid and quantitative
models of road user decision making. Elementary utility models (O'Neill, 1977;
Janssen and Tenkink, 1988) have already paid some services in this respect. In the
Janssen and Tenkink model (see Fig. 8.1), the road user is assumed to balance the
(dis)utilities of time loss during the trip, plus the possible accident risk, against
the utility of being at the destination. From this a choice of optimal speed, and
possibly of other driving behaviour parameters, then follows so as to be at the
optimum of that balance.
It has been derived, for example, from this type of consideration that a device
having an expected effectiveness (i.e. an engineering estimate) 8 will not reduce
accident risk with that same factor but with a factor that happens to be

== 1 - (1 - 8)-1/(c+l) , (1)

where c is Nilsson's (Nilsson, 1984) parameter in his speed-risk function, with

values between 3 and 7 for different types of accidents. For fatalities, c == 7.
It is clear that the safety effect to be realised will always be less than the ex-
pected effectiveness (see Fig. 8.2). For example, if we take the commonly used
value of 8 == 0.43 for the seat belt, the estimated effect to be achieved for the
fatality rate per km would be in the order of 7% rather than 43% (at 100%
use rate).
 8. Modelling Driver's Risk Taking Behaviour 139

occident loss, after

time loss
loss accident loss, befor~
per km

- - - total loss. otter

- ...
,
" total loss, before

o,....-~-------+---+------v

- resulting occident loss,

before & after
loss
per minute

occident loss, after

---------------+-- time loss

accident loss. before

FIGURE 8.1. Utility model (Janssen and Tenkink, 1988) shows how drivers select optimum
speed as a function of time (opportunity) losses and accident risk so as to make the resulting
total expected loss minimal. It appears to be generally true that whenever accident risk is
objectively reduced ('after' situation) the optimum speed that is selected will move towards
the higher end of the scale.

8.4.2 A Word of Caution About Working with Risk Measures

in Traffic Safety Studies
In the risk sciences, it is good to remind ourselves from time to time what is the
'risk' that we are dealing with. In traffic safety, for example, risk can be expressed
per kilometre travelled, per capita/vehicle (per year) or per unit of time spent in
140 Janssen

1.0 FIGURE 8.2. Expected and actual (i.e.

pre- or postdicted) safety benefit, ac-
cording to a simple utility model of
0.8 driver behaviour.
<W
en
(12
Q)
c 0.6
eu
>
.~

0
~
Q) 0.4
ii
:1
.....
U
co
0.2

0.0 0.2 0.4 0.6 0.8 1.0

effectiveness E of safety device

traffic. It should be realised that these indicators can go in different directions at

the same time and this has led to much confusion, e.g. in criticising Wilde's Risk
Homeostasis Theory or variations thereof. Likewise, the finding of some risk level
being not constant from one period of time to the next is not necessarily a valid
criticism of these formulations.

8.4.3 A Piece of Empirical Evidence from

Seat Belt Accident Statistics
Hardly does one ever have access to all components in the chain. However, there
appears to be one case in which almost all is available, which happens to be the
safety belt.
The statistics for this case comprise a set of data from the Federal Republic of
Germany pertaining to a sudden rise in seat belt wearing rate and its subsequent
effect on passenger car driver fatalities (Briihning et al., 1986).
From 1 August 1984, onward German road authorities exerted a strict en-
forcement of seat belt legislation by setting a fine of 40 Deutschmark ('Verwar-
nungsgeld') for being apprehended as a non-wearer passenger car driver. Almost
overnight, the overall wearing rate in the country rose from 58 to 92%. This makes
the German data as close as coming from an ideal 'natural' experiment as pos-
sible and it makes it feasible to postdict fatalities after the increase in wearing
rates, given the availability of an engineering estimate of the belt's effectiveness
in preventing a fatality when a crash happens.
Obviously, the spectacular increase in wearing rate should be reflected in an
immediate downward step in passenger car fatalities. As a matter of fact, depending
on the engineering estimate used for the seat belt's intrinsic effect the reduction
in fatalities should have been between 24% and 31% (that is, the increase in use
 8. Modelling Driver's Risk Taking Behaviour 141

rate multiplied by the engineering estimate). On the other hand the utility model,
on the basis of Equation (1), would have predicted a reduction of between 3% and
4%. The actual reduction was 6.7%. The readers may draw their own conclusions
as to which of the two was the best prediction (or rather, postdiction): Janssen,
(1994b), has further discussion 1.

8.4.4 Higher-Order Forms ofAdaptation

Following are the other forms of adaptation that may occur as the result of having
a new support available:

The generation of extra mobility (VMT). For example, navigation systems may
not so much reduce excess mileage as generate extra mileage into areas that
were formerly avoided. Or entrepreneurs who formerly 'lost' 5% or 6% of the
mileage driven by their fleet because drivers selected non-optimal routes to their
destination may now plan an extra trip a day because navigation performance
has become flawless.
Road use by less qualified segments ofthe driving population. It is to be expected
that some categories of users that did not dare to venture out in traffic, realis-
ing their own imperfections, will do so if offered an extra amount of 'built-in'
safety.
Driving under more difficult conditions. Similarly, the extra safety offered will
tempt road users to move to places they formerly avoided.

All these effects lend themselves to modelling by elementary utility considerations,

as discussed earlier.
On the level of aggregate accident statistics, a negative correlation between fatal
accident rates and VMT is predicted by utility considerations and has often been
observed. For example, a British investigation (Shannon, 1986) showed that for
the period from 1973 to 1983, there was a correlation of -0.88 between the annual
fatal accident rates (per mile) and VMT on British motorways. Exactly the same
value can be calculated from that paper for US Interstate data. Finally, in Japan
the death rate per km driven fell on average by a factor of 1.12 per year between
1966 and 1982, while the motorised mileage rose by an average of 1.08 from year
to year in that same period. The correlation between these two rates amounted
to -0.97. Thus, those years that were marked by relatively large decreases in the
death rate per km were also marked by relatively large increases in kilometre per
capita.
On a still higher level, many authors have noted that what is good for macro-
economy is bad for traffic safety and vice versa and that this is not just because of
economy-induced fluctuations in VMT (Joksch, 1984; Partyka, 1984; Wagenaar,
1984; Wilde, 1991). Table 8.2 lists correlations between annual unemployment

1 In which it is shown that selective recruitment is not an explanation of the less-than

expected safety effect in the German data.
142 Janssen

TABLE 8.2. Correlation between annual

unemployment rates and same-year per capita
traffic death rates (from Wilde, 1991).
United States, 1948-1987 -0.68
Sweden, 1962-1987 -0.69
West Germany, 1960-1983 -0.83
Finland, 1965-1983 -0.86
Canada, 1960-1986 -0.86
United Kingdom, 1960-1985 -0.88
The Netherlands, 1968-1986 -0.88

rates and same-year per capita traffic death rates in seven Western countries. Taken
together, the data do appear to confirm the basic assumption that safety and utility
are related to each other, i.e. safety is a factor in the utility considerations associated
with undertaking a trip.

8.5 The Link Between Behaviour and Risk

It would be of great help to the prediction of net safety effects if, given certain en-
gineering estimates, use rates, behavioural changes, etc., the relationships between
parameters conventionally used to describe driving behaviour and accident risk are
known. This would then permit the translation of effects observed in behavioural
studies in simulators and instrumented vehicles into safety effects to be expected
in the population. While the issues are still not resolved, progress has indeed been
made recently with respect to several important parameters.

8.5.1 Average Speed, Speed Variability and Risk

There presently exist several models that relate average speed and/or speed vari-
ability to accident risk. The most pragmatic and useful one is the one by Nilsson
(1984), who distinguishes between different types of accident (fatal; killed + se-
riously injured; material damage). The functions are power functions of average
driving speed, with powers ranging from three (material damage only) to seven
(fatal). With respect to speed variability and risk functions, useful and ready-to-use
expressions have been derived by Salusjarvi (1990).

8.5.2 Lane-Keeping Performance and Risk

The situation is somewhat less clear-cut for lateral performance indicators. It seems
that for the time being we have to be satisfied with proxies like TLC and the
frequency with which lane exceedances occur.
 8. Modelling Driver's Risk Taking Behaviour 143

8.5.3 Car-Following and Risk

In the seminal work of Farber (1993, 1994), a set of car-following data measured
in actual traffic was used to assess the impact of a collision-avoidance system
that would effectively reduce the following driver's reaction time to a sudden
braking action by the preceding vehicle. This can be generalised to calculating
the risk attached to a given situation per see The algorithm has the following steps
(Janssen, 2000):

For a given headway it is calculated whether, for a given range of reaction times,
a collision would follow if the preceding vehicle were to brake vehemently, i.e.
at full braking power.
The total probability of a collision is then computed by integration over a log-
normal distribution (which has a tail towards the longer reaction times) of driver
reaction times.
The mean and the standard deviation of the distribution are, moreover, adapted
to headway itself. This procedure was introduced by Farber so as to indicate that
drivers follow more attentively at shorter headways.
In case of a rear-end collision, the speed difference at the moment of impact is
computed. The overall risk of the car-following situation is then computed by
multiplying accident probability by the squared speed difference at impact.

Fig. 8.3 illustrates results for a few everyday car-following situations. As has
been observed by other authors, there is a 'worst' headway to follow, which is not
at the shortest range. This is intuitively clear when it is realised that although the
probability of the collision itself happening becomes higher at shorter headways,
its severity will be less because at a short headway the speed difference between
the two vehicles at the moment of impact will be less.

300 v1 =20 m/s

dec. =-8 m/s2
250

200

~ 150
a:
0

100

50

0
0.25 0.5 0.75 1 1.25 1.5 1.75 2
Headway

FIGURE 8.3. Rear-end collision risk when following a vehicle at a certain headway that
drives at 20 m/s and brakes suddenly (for two speeds of the following vehicle). Risk units
are arbitrary, i.e. defined as 100 at one of the configurations.
144 Janssen

8.6 Countermeasures Against Behavioural Adaptation

8.6.1 Should There Be Any?

The first thing to be kept in mind when pondering the question of what to do
against behavioural adaptation is that there is a good side to it as well. One has
better primary performance at a somewhat decreased accident risk. Likewise, the
extra mobility that is generated (grandmother gets out on the road again) is an
asset, i.e. a social gain.

8.6.2 Incentive Schemes and Their Expected Results

To move the optimal speed that drivers select, according to the utility model, in
one direction or the other, costs and benefits of driving in a more or less risky way
should be changed. This is the theoretical way out and empirical evidence from lab
studies and others indicates that this approach can work. The model then permits
to predict what safety effects to expect from these so-called incentive schemes
(e.g. Janssen, 1990).

8.7 Conclusions

There is now at least some useful knowledge available on the behavioural effects of
driver supports, which permits more than an educated guess on what safety effects
will follow from the provision of these supports. This derives both from a more
advanced insight into the behaviour itself and from the availability of procedures
to translate behavioural effects into safety effects. Thus

behavioural adaptation is definitely here to stay;

we better start modelling it, preferably in a quantitative way;
we should forget our prejudices against econometric principles, because people
are not mental cripples, as many (cognitive) psychologists want us to believe.

8.8 An Afterthought

When driver support systems that offer a safety benefit are introduced on the road
their benefits will, in all likelihood, be less than what originally could be expected.
This is, by itself, an important fact of life. However, as some authors have surmised,
the introduction of any specific safety measure could well be no more than a tiny
bubble on top of a continuously ongoing societalleaming curve of a much broader
nature. Although Smeed's ideas (Smeed, 1949, 1968, 1972; Smeed and Jeffcoate,
1970) about the learning process that comes along with a society's increasing level
of motorisation are no longer popular today, they may well reflect what is the more
significant permanent background against which all incidental safety measures
 8. Modelling Driver's Risk Taking Behaviour 145

become relatively minor. Modelling that background process is maybe really what
we should aim for in the long term.

References
Aschenbrenner, K., Biehl, B. and Wurm, G. (1994). Mehr Verkehrssicherheit durch bessere
Technik? Felduntersuchungen zur Risikokompensation am Beispiel des Antiblockiersys-
terns (ABS). Bast, Bergisch Gladbach, Bericht 8323.
Bruhning, E., Ernst, R., Glaeser, H.P., Hundhausen, G., Klockner, J.H. and Pfafferott, I.
(1986). Zum Riickgang der Getotetenzahlen im StraBenverkehr-Entwicklung in der
Bundesrepublik Deutschland von 1970 bis 1984. Zeitschrift fiir verkehrssicherheit, 32,
154-163.
Evans, L. (1985). Human behaviour feedback and traffic safety. Human Factors, 27, 555-
576.
Farber, E. (1993). Using freeway traffic data to estimate the effectiveness of rear-end colli-
sion countermeasures. Proceedings ofthe Third Annual Meeting ofthe Intelligent Vehicle
Society ofAmerica, Washington, DC.
Farber, E. (1994). Using the Reamacs model to compare the effectiveness of alternative rear
end collision warning algorithms. XIVth International Technical Conference on Enhanced
Safety of Vehicles, Miinchen. Paper Nr. 94 S3 0 03.
Janssen, W.H. (1988). Gurtanlegequoten und Kfz-Insassen-Sicherheit: eine Anmerkung zu
jiingsten deutschen Erkentnissen. Ze itsch rift fiir verkehrssicherheit. 34, 65-67.
Janssen, W.H. (1990). The economy of risk; or, what do I get for my error? Ergonomics,
33,1333-1348.
Janssen, W.H. (1994a). Seat-belt wearing and driving behaviour: An instrumented-vehicle
study. Accident Analysis and Prevention, 26, 249-261.
Janssen, W.H. (1994b). Behavioural adaptation to road safety measures: A framework and an
illustration. In R.M. Trimpop and GJ.S. Wilde (Eds.). Challenges to Accident Prevention.
Styx, Groningen, pp. 91-100.
Janssen, W.H. (2000). Functions relating driver behaviour and accident risk. Report TM-
00-D004. TNO Human Factors, Soesterberg, The Netherlands.
Janssen, W.H. (2001). Advanced driver supports, driver behaviour, and road safety. Pro-
ceedings ofthe 8th World Congress on Intelligent Transport Systems, Sydney. (CD-ROM)
Janssen, W.H., Kaptein, N. and Claessens, M. (1999). Behaviour and safety when driving
with in-vehicle devices that provide real-time traffic information. Proceedings of the 6th
World Congress on Intelligent Transport Systems, Toronto. (CD-ROM)
Janssen, W.H. and Tenkink, E. (1988). Considerations on speed selection and risk home-
ostasis in driving. Accident Analysis and Prevention, 20, 137-143.
Joksch, H.C. (1984). The relation between motor vehicle accidents deaths and economic
activity. Accident Analysis and Prevention, 16,207-210.
Nilsson ,G. (1984). Speeds, accident rates and personal injury consequences for different
road types. Report number 277VTI, Linkoping, Sweden.
O'Neill, B. (1977). A decision-theory model of danger compensation. Accident Analysis
and Prevention, 9, 157-165.
Partyka, S.C. (1984). Simple models of fatality trends using employment and population
data. Accident Analysis and Prevention, 16, 211-222.
Salusjarvi, M. (1990). Finland. In G. Nilsson (Ed.). Speed and Safety: Research Results
from the Nordic Countries. Linkoping.
146 Janssen

Shannon, H.S. (1986). Road-accident data: Interpreting the British experience with partic-
ular reference to the risk homeostasis theory. Ergonomics, 29(8), 1005-1015.
Smeed, R.I. (1949). Some statistical aspects of road safety research. Journal of the Royal
Statistical Society, Series A, Part I, 1-34.
Smeed, R.I. (1968). Variation in the pattern of accident rates in different countries and their
causes. Traffic Engineering and Control, 364-371.
Smeed, R.I. (1972). The usefulness of formulae in traffic engineering and road safety.
Accident Analysis and Prevention, 4, 303-312.
Smeed, R.I. and Jeffcoate, G.O. (1970). Effects of changes in motorisation in various
countries on the number of road fatalities. Traffic Engineering and Control, 150-151.
Wagenaar, A.C. (1984). Effects of macro-economic conditions on the incidence of motor
vehicle accidents.Accident Analysis and Prevention, 16, 191-205.
Wilde,G.I.S. (1991). Economics and accidents: A commentary. Journal of Applied Be-
haviour Analysis, 24, 81-84.
9
Dealing with Behavioural Adaptations
to Advanced Driver Support Systems
FARIDA SAAD

9.1 Introduction
Over the past 15 years major technological advances have been made in the field of
automotive technology. Many research and development programmes (in Europe,
Japan and the United States) have been devoted to the design and implementation
of new driver support systems and information management systems (for route
planning, obstacle detection, car-following situations, speed control and so on).
The development of these systems raises crucial issues at a technical level as
well as in terms of their consequences on driver activity (for an overview, see,
e.g., Michon, 1993; Parkes and Franzen, 1993; Noy, 1997). Some of these issues
deal, in particular, with the conditions of use of the systems, their effects on driver
behaviour and strategies and their impact on the operation and safety of the traffic
system. A major concern is about the 'behavioural adaptations' that may occur in
response to the introduction of these systems in the driving task and their impacts
on road safety (Smiley, 2000).
These new support systems will mediate drivers' interactions with their driving
environment (vehicle, road infrastructure, traffic rules, other road users) by creating
new sources of information and/or offering new modes of vehicle control. They
will thus alter the conditions in which the driving task is currently performed and
can thus be expected to engender changes in drivers' activity. Changes may occur
(1) within the very activity of 'supported' drivers (in terms of divided attention
between the new internal sources of information and direct monitoring of the road
environment, changes of driving strategies, delegation of control to the driving
support system and so on); (2) within their interactions with other road users (effect
on the behaviour of other road users, 'readability' of assisted drivers' behaviour for
other drivers, etc.). It is then important to specify the nature, direction and extent
of the changes likely to occur at these different levels, since they will determine
the ultimate impact on road safety (Evans, 1985; OCDE, 1990).
The changes associated with the use of these new support systems and their
acceptance by drivers will depend (1) on the types of task they are designed
to support (navigation, guidance or control tasks; Allen et al., 1971); (2) their
functions and the type of mediation they provide ('description' as regards the state

147
148 Saad

of the driving environment, 'prescription' as regards the regulating action the driver
has to take; 'intervention' or 'taking over' part of the driving task in the event of
deliberate driver delegation or of driver failure).
Up to now, most support systems are dedicated to specific driving tasks. Their
competence is by definition limited to the area of that task (or a subset of conditions
in which that task is performed, such as conditions of good visibility for instance).
The mediation offered is thus only partial, the driver's direct control over the
driving environment is always necessary and he remains responsible for the overall
management of his journey.
Studying the integration of the systems into the overall driving task and iden-
tifying behavioural changes when using them are thus critical aspects that need
to be carefully studied and analysed. This entails (Saad and Villame, 1999) the
following:
Taking account of the essential dimensions of the road environment in which
that activity takes place (nature of the interactions at work, regulatory, structural
and dynamic constraints, etc.). This reference to the context (Suchman, 1987)
is particularly important in view of the diversity and variability of the road
situations that drivers may encounter during a journey.
Choosing functional units of analysis making it possible to examine not only
the impact on the performance of the specific task to which the support system
is dedicated (compliance with safety margins or speed limits, for instance), but
also its compatibility with the performance of other driving tasks (overtaking,
interacting with other users, etc.).
Selecting the relevant indicators for revealing the changes likely to take place in
driver's activity.
These issues make direct demands on our knowledge of the driving task and of the
psychological processes (cognitive and motivational) that govern drivers' activity.

9.2 'Behavioural Adaptation' in Road Safety Research

In road safety research, the term 'behavioural adaptation' is mainly associated
with unintended or unexpected behavioural changes that may appear in response
to the introduction of a change in the traffic system and which may (more or less)
jeopardise road safety. Thus, the emphasis is placed primarily on the negative
aspects of the phenomenon.
For example, an OECD expert group (1990) defined behavioural adaptations
as 'those behaviours which may occur following the introduction of changes to
the road-vehicle-user system and which were not intended by the initiators of
the change'. On the basis of a review of a large number of empirical studies,
the expert group concluded that 'behavioural adaptation does occur, although not
consistently'. Behavioural adaptation may be an immediate response to the change
introduced in the traffic system or may only appear after a long time. Generally,
behavioural adaptation does not eliminate safety gains from measures, but tends
to reduce the size of the expected safety effects. Different elements are assumed
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 149

to influence the occurrence of behavioural adaptation such as the nature and the
'perceptibility' of the changes introduced in the traffic system (changes that directly
influence the way the driving task is performed or changes that alter the driver's
subjective safety, for instance), the degree of freedom that the change allows drivers
(changes that give the driver an opportunity for adapting his behaviour) or the
presence of competitive motives (safety versus mobility or productivity motives,
for instance).
Although behavioural adaptation is a widely acknowledged phenomenon, the
factors likely to explain it and the processes underlying its occurrence are not
clearly established. Numerous processes may in fact come into play between the
introduction of an 'innovation' in the traffic system and its 'adoption' by drivers,
its 'translation' into behaviour (whether 'safe' or 'risky') and its longer term con-
sequences on the operation and safety of the traffic system (Brown, 1985).
These processes should be analysed in greater depth, in particular those in-
fluencing the way drivers interact with their driving environment (vehicle, road
infrastructure, traffic rules, other road users, etc.). Such analyses should help to
formulate hypotheses about the changes in behaviour that may occur, identify the
conditions in which a 'negative' compensation for safety might appear and direct
thoughts on the means of minimising the extent of such negative changes.
Within the European project adaptive integrated driver-vehicle interface
(AIDE 1) , which aims to develop an harmonised interface that integrates all in-
vehicle support and information systems, a research activity is devoted specifi-
cally to identifying crucial behavioural adaptation issues associated with the use
of new support systems and determining the most relevant parameters that can be
implemented in models for supporting design and safety assessment processes.
The aim of this paper is to present the main phases of this research activity and
the results obtained so far. We begin by outlining the main results of a literature
review on 'behavioural adaptation' to new driver support systems (Saad et al.,
2004), especially advanced driver assistance systems (ADAS), which intervene
more or less directly in the performance of the driving task, such as adaptive cruise
control (ACC) or intelligent speed adaptation (ISA). We then describe the ongoing
activity associated with the conduct of experimental studies in order to improve
our knowledge on short and long term behavioural adaptation (Saad et aI., 2005)
and to develop models that can act as a reference for the design of an adaptive,
integrated in-vehicle interface supporting the multiple tasks drivers have to perform
in modem vehicles (Cacciabue and Hollnagel, 2005).

9.3 Behavioural Adaptation to Advanced Driver

Support Systems
In most research on 'behavioural adaptation' to new support systems, one of the ma-
jor concerns is the identification of 'adverse behavioural consequences' (Grayson,

1 AIDE is an EC Funded Project of the 6th Framework Programme - Project N. IST-l-

507674-IP
150 Saad

1996) or negative side effects associated with their use and which may reduce the
expected (safety) benefits of the assistance provided to the driver. In many research
studies, it is assumed that driving with systems which take over some elements
of the driving task (such as speed and time headway control) may reduce drivers'
workload and provide them with an opportunity for devoting less attention to the
(primary) driving task. Another concern relates to the drivers' ability to cope with
the limitation of support systems and to resume control in 'safety critical' traffic
scenarios, either because of drivers' misconceptions about the functioning of the
system and over-reliance on the system or as a consequence of drivers' reduced
attention to the driving task. In some studies, particular emphasis is placed on the
possible deterioration of drivers' interactions with other road users.
These research orientations guide the choice of the driving performance indica-
tors taken into account for assessing behavioural changes (such as driving speed,
safety margins in car-following situations and lateral control of the vehicle, per-
formance to a secondary task or subjective assessment of workload). They also
guide the choice of the driving situations or scenarios examined ('normal' driving
situations, 'safety critical' driving situations such as take-over situations in which
the driver has to regain control of his vehicle because of the limitations themselves
of the systems or a technical failure).
The second major concern relates to drivers' opinions and their acceptance of
the assistance provided. In most research studies, perceived usefulness of and sat-
isfaction about the systems are assessed either through standardised questionnaires
(see, e.g., Van der Laan et al., 1997) or through drivers' verbal reports and in-depth
interviews. In some studies, drivers' acceptance of the support system is more pre-
cisely assessed through the very usage of the system (in terms of drivers' decision
to engage the systems in various situational contexts and the overall duration of
system engagement, for instance).
Several empirical studies have been carried out, focusing mainly on the impact
of individual support systems, such as collision avoidance systems (CAS), speed
limiters, ISA or ACC systems, either in the' controlled' context of driving simulator
or in the complexity of real driving situations. Most of these studies have been
short-term studies and 'the effects (of support systems) on traffic safety and driver
behaviour are still uncertain in many respects' (Nilsson et al., 2002), especially their
long-term effects. Nevertheless, some critical issues have already been identified.
These issues are presented and discussed below.

9.3.1 The Diversity ofBehavioural Changes Studied

and Observed
The first critical issue encountered when examining the impact of a given support
system concerns the diversity of behavioural changes studied and observed, as
well as the magnitude and direction of these changes.
For example, the main behavioural changes observed when studying the be-
havioural impact of ACC are changes in speed, in the safety margins (time headway
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 151

or time-to-collision) adopted in various car-following situations (such as steady

car-following, catching up a slower vehicle, etc.) and in the lateral control of the
vehicle, as well as changes in lane occupancy and in the frequency of lane change
manoeuvres. The results obtained are sometimes contradictory. For example, in
some studies, the average driving speed increased when using ACC (Ward et al.,
1995; Hoedemaeker and Brookhuis, 1998), whereas in others this was not the case
(Hogema and Janssen, 1996; Stanton et al., 1997; Tornes et al., 2002). The same
divergent tendencies were observed when it came to the frequency of unsafe safety
margins adopted in car-following situations and the lateral position of the vehi-
cle when using ACC. For example, Fancher et al. (1998) and Saad and Villame
(1996) observed that the use of ACC reduced the frequency of short-time head-
way, Stanton et al. (1997) and Hoedemaeker and Kopf (2001) found no differences
between manual and ACC driving and Ward et al. (1995) observed a tendency to
drive with shorter time headways with ACC. Sometimes the results are more con-
vergent. For example, when driving with ACC on motorways, one observes a
general tendency for drivers to spend more time in the left lane (Hoedemaeker and
Brookhuis, 1998; Nilsson, 1995; Saad and Villame, 1996; Tornros et al., 2002).
In 'safety critical' scenarios, which require the drivers to reduce their speed or
to brake, drivers generally react later and/or with reduced safety margins when
driving with ACC (Nilsson and Nabo, 1996; Hoedemaeker and Brookhuis, 1998;
Rudin-Brown and Parker, 2004).
The same divergent trends are observed when studying the behavioural impact
of ISA. For example, changes in car-following behaviour were observed when
driving with ISA and the direction and magnitude of these changes varied from
one study to another. In urban areas, Varhelyi and Makinen (2001) and Hjalmdahl
and Varhelyi (2004) found that, when driving with ISA, the time headway in car-
following situations increased, whereas Carsten and Fowkes (2000) observed an
increase in close following (time headway less than 1 s). On rural roads, the results
are more convergent and suggest that there is an increase in close following.
As regards the workload associated with the use of the systems, the results also
reveal some variations. Nilsson (1995) and Ward et al. (1995) found no difference
in subjective workload between usual driving and ACC driving, while Hoede-
maeker and Brookhuis (1998) and Tornros et al. (2002) found that the subjective
workload was lower in ACC driving. Stanton et al. (1997) and Rudin-Brown and
Parker (2004) found that a secondary task was better performed when driving with
ACC, while this was not the case in the study of Young and Stanton (2002). Some
studies suggest that workload depends on the characteristics of the systems stud-
ied and/or the task to be carried out (Nilsson and Nabo, 1996; Hoedemaker and
Kopf, 2002; Young and Stanton, 2004). Other studies suggest that workload also
depends on drivers' degree of familiarisation with the system (Kopf and Nirschl,
1997).
Thus, part of the observed diversity may be due to the functional characteristics
of the systems studied (for ACC, control algorithm, time headway targets avail-
able, deceleration level used, etc.) and to the degree of drivers' experience with
them.
152 Saad

More generally, the diversity of the results obtained raises questions about the
methods used, the type and number of variables selected for assessing the im-
pact of the system, and finally the (implicit or explicit) models governing their
choice. When examining the results, attention should be paid to these theoretical
and methodological issues. In particular, the context in which the studies have
been carried out (driving simulator, closed tracks or real driving situations) should
be specified as well as the various scenarios and driving tasks in which the be-
havioural changes have been identified. The diversity of the results obtained could
then be examined and discussed in the light of the characteristics of these vari-
ous contexts and scenarios. Such analysis is particularly relevant insofar as the
situational context plays an important role in the behavioural changes observed
when driving with a support system and more generally in drivers' activity (Saad,
2002).

9.3.2 The Importance of the Situational Context

and the Interactive Dimension ofDriving
The second critical issue when studying behavioural changes when driving with
ADAS is related to the diversity and variability of the road situations that drivers
may encounter during a journey.
Many systems are designed to support drivers in maintaining some safety thresh-
olds or ensuring compliance with some formal driving rules (such as maintaining
safe time headways in car-following situations or adhering to legal speed limits),
independently of the characteristics of road situations (infrastructure and traffic
related) and the task planned or being performed, which determine the driver's
current regulating actions.
Several studies reveal the influence ofthe road infrastructure and traffic condi-
tions on the driver's decision to use the support systems (decision to engage ACC
and ISA systems or a speed limiter; Comte, 2000; Fancher et aI., 1998; Saad and
Malaterre, 1982) and his willingness to follow the systems' 'advices' (Malaterre
and Saad, 1986) as well as on the magnitude of the behavioural changes observed
when using them (for instance, increased safety margins before overtaking only in
light traffic conditions when receiving time headway feedback - Fairclough et al.
(1997) or when driving with an ACC - Saad and Villame (1996).
These results suggest that drivers' use and acceptance of the systems closely
depend on the way they integrate (safety) formal rules in their driving and the tol-
erances they deem admissible, according to the situational context (infrastructure
and/or traffic related) and the task planned or carried out ('stable' car-following,
pulling in or pulling out manoeuvres, etc.).
Drivers' use and acceptance of the assistance provided also depend on its im-
pact on the way they usually manage their interactions with other drivers (on the
basis of more or less informal rules or behavioural norms; Saad et aI., 1999). In
many interaction situations, such as driving in dense or unstable traffic condi-
tions, drivers seem reluctant to use the systems when doing so would require a
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 153

significant deviation from their usual strategies. For example, in a field trial with
a 'Driver Select' ISA (one which the driver can choose to engage or not), Carsten
and Fowkes (2000) observed that drivers were prone to disengage the system in
areas where speeding was the norm for the surrounding traffic. In such traffic
conditions, drivers prefer to be in control of their speed and tum the system off
when they feel under pressure from other drivers. Furthermore, drivers are con-
cerned about the way other drivers might interpret their own behaviour (Saad and
Malaterre, 1982). Certain aggressive reactions on the part of other drivers (close-
following behind, cutting-in manoeuvres, flashing headlights, etc.) are perceived
as negative social feedback and often lead them to give up the use of the support
system.
Finally, some studies have shown that driving with ISA may also influence the
driver's interaction with other road-users (at junctions or at pedestrian crossings in
urban areas), either negatively (Persson et aI., 1993) or positively (Almqvist and
Nygard. 1997) in the short term, but with the probability of improvement after
longer experience with the system (Hjalmdahl and Varhelyi, 2004).
These studies highlight the circumstantial requirements of driving assistance
according to the dynamics of various environmental conditions and to the drivers'
motives, objectives and intentions in these conditions. They also confirm the need
to adopt a multi-level approach when assessing behavioural adaptation to new
driver support system, which is to say to study possible changes within the activity
of 'assisted' drivers as well as within their interactions with other road users.

9.3.3 The Potential Differential Impact ofDriver

Support Systems
The third critical issue encountered when studying behavioural adaptation is related
to the potential differential impact of support systems. The question at issue here
is to find out whether some categories of drivers are more prone to adapt their
behaviour than others when driving with new support systems (see, e.g., Jonah,
2001).
Because of the great diversity of the driver population (in terms of both car us-
age and individual characteristics), many driver characteristics may be considered
relevant for dealing with this issue, such as drivers' age and gender, their degree
of driving experience and practice, their degree of familiarity with new technolo-
gies, their attitudes towards driving and traffic rules and so on. Choosing a set of
individual characteristics to take into account depends primarily on the objectives
of the research study and the processes under investigation.
In some studies, the concept of 'driving style' has received particular attention. It
is not in the scope of this short presentation to discuss either the various dimensions
characterising 'driving style' or the different behavioural indicators used to render
this variable operational. Basically, 'driving style' is described as a relatively stable
characteristic of the driver, which typifies his or her personal way of driving, the
way he or she chooses to drive (for instance, the level of speed or the safety margins
154 Saad

most frequently adopted, the level of attention devoted to the driving task and so
on; French et al., 1993). Several studies have taken into account this variable when
studying the impact of ACC, whether by design (Hoedemaeker and Brookhuis,
1998, who selected the participants on the basis of two dimensions of the driving
style questionnaire established by French et al. op.cit., namely 'speed' and 'focus ')
or a posteriori, on the basis of the identification of some manifest behaviour patterns
(such as the driver's propensity to change lane frequently on the motorway - Saad
and Villame, 1996 - or the driver's tendency to drive faster or slower than the
surrounding traffic and to adopt short-time headways in car-following situations;
Fancher et al., 1998). The results suggest that the various dimensions of 'driving
style' taken into account in these studies may play an important role in the use and
acceptance of new driver support systems and in the occurrence of behavioural
changes when using them. For example, some behavioural changes observed in
ACC driving, such as a reduction in the number of lane change manoeuvres and
a higher rate of left-lane occupancy, were primarily observed within the group of
drivers who usually tend to change lane frequently when driving on motorways,
while no significant changes were observed for the other" rather less mobile" group
(Saad and Villame, Ope cit.). Fancher et al. (1998) also suggest that driving style
may account for the differences in the overall use of ACC. A particular finding was
that drivers who are described as " hunter/Tailgater " because they drive fast and are
inclined to adopt short-time headways used ACC less often than the other groups.
Hoedemaeker and Brookhuis (1998) also identified some differences as regards
(self-reported) driving styles, depending on the variable taken into consideration.
For instance, while all drivers increase their driving speed with ACC, irrespective
of their driving style, in a critical scenario (in this case a situation in which the
driver has to brake in response to the full stop of a lead vehicle) 'low speed' drivers
braked harder when driving with ACC while 'high speed' drivers braked the same
as when driving without ACC. Differences in driving styles have also an effect
on drivers' acceptance of the assistance provided: 'high speed' drivers are less
positive about ACC than 'low speed' drivers.
Other authors (Rudin-Brown and Parker, 2004; Ward et al., 1995) took into
account some general personality traits, namely 'Sensation Seeking' and/or 'Locus
of Control' (LOC) when studying the effect of ACC on driver behaviour. These
personality traits are assumed to influence, more or less directly, the occurrence of
behavioural adaptation either through a general propensity for 'risk compensation'
(for 'high sensation seekers ') or a tendency to manifest over-reliance on automation
(for'external LOC'). Their results suggest that these individual characteristics tend
to amplify some behavioural changes observed when driving with ACC, such as
impaired lane keeping, slower reaction to the activation of the brake lights of a
lead vehicle (more pronounced for 'high sensation seekers' than for 'low sensation
seekers') or slower reaction to a (simulated) failure of ACC (more pronounced
for drivers with external LOC than for drivers with internal LOC). Individual
characteristics also influence drivers' subjective assessments of the impact of the
system on their driving (for instance, 'high sensation seekers' reported lower level
of arousal and effort when driving with ACC than 'low sensation seekers ').
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 155

The results suggest that individual driver characteristics, such as 'driving style' ,
'sensation seeking' or 'locus of control' , seem to playa role in the overall frequency
of ACC usage, in the occurrence and/or the magnitude of some behavioural changes
when using it and in the acceptance of the assistance provided. It should be noted,
however, that the relationship between general personality traits and other individ-
ual characteristics such as 'driving style' are not quite clear and should be more
precisely established.

9.3.4 Learning to Drive with New Driver Support Systems

As emphasised in the introduction, most support systems are dedicated to a specific
driving task and their function is by definition limited to the area of that task. It
is important therefore to determine whether drivers can easily learn the scope and
limits of the support system's competence. This issue is of vital importance for
support systems that intervene directly in vehicle control.
The learning process is certainly crucial for helping drivers to build an
appropriate representation of the assistance provided by the system and for
'calibrating' their trust in it. Appropriate mental models of and confidence in new
driver support systems should promote their optimal (and safe) use by drivers
(Muir, 1987; Amalberti, 1996). However, to our knowledge, these aspects have
received little attention up to now and only a few research studies dealing directly
with this issue have been identified. These studies dealt primarily with the learning
process of ACC systems.
Some research has covered the learning issue by using questionnaires and in-
terviews to try and gauge the ease with which drivers think they have learned to
use ACC and the elements that facilitated the learning process. Nilsson (1995)
indicates that the drivers rated the ACC system very easy to learn and to ma-
noeuvre. According to Faucher et al. (1998), ACC was found to be 'rather easy
to use, quick to learn, satisfying in its use and more or less straightforward to
supervise in the hands of most lay drivers'. An interesting point emphasised in
their study is that the kinaesthetic feedback provided by the action of ACC seems
to be a primary factor determining the driver's relative ease in learning to use the
system.
Other studies have investigated the learning process of ACC systems in greater
detail and especially in 'take-over' situations, where the driver has to decide
whether to regain control over his speed or not. Kopf and Nirschl (1997) studied
the learning of three versions of ACC (differing mainly in the maximum decel-
eration which could be applied by the system - soft, medium and hard - and the
set values for the time headway). The participants performed five 130-km jour-
neys with ACC. The findings indicate that the frequency of drivers' interventions
and the (subjective) workload decrease as drivers become more experienced. They
also reveal that a different layout of ACC parameters influences driver behaviour
with respect to learning and workload. An in-depth analysis of long-term driver
interactions with an ACC was carried out by Kopf and Simon (2001). The partici-
pants drove with ACC for 2.5 weeks. The results indicate that ACC usage evolves
156 Saad

through different stages: a preliminary stage of getting to know the system (learn-
ing to operate it); a testing stage (learning the system limits); and a familiarisation
stage (learning to use the system appropriately according to the situational context).
An analysis of changes over time in drivers intervention in 'take-over' situations
suggests that there was a trend towards testing the limits of the system at the be-
ginning, followed by a certain apprehension of the system's capabilities and then a
more personalised 'steady use' of the system. Weinberger et al. (2001) conducted
a long-term operational field test to obtain more information about the learning
process as regards both the usage of controls and display and the judgement ap-
plied in take-over situations. The participants used an ACC-equipped car over a
period of 4 weeks each. Both drivers' self-assessment of the length of the learn-
ing process and drivers' behaviour during 'take-over' situations suggest that 2 or
3 weeks are needed to learn the operation of ACC and the assessment of take-over
situations. However, as the participants in the study drove much more than the av-
erage driver, the authors suggest that other ACC users might need a longer learning
time.
These studies provide some useful insights into the duration of the learning
process of ACC and its different stages and into the way in which drivers' in-
teractions with the system change and develop over time. They also suggest that
it is important to find means to accelerate the learning process, by supporting
drivers' predictive activities about the behaviour of ACC and more generally by
helping drivers develop appropriate conceptions of the systems' behaviour and
limits. More generally, these studies highlight an important dimension to take
into account when dealing with behavioural adaptation, the temporal dimension
of behavioural changes.
It should be noted that the issue of learning to use new support systems is
attracting more attention, as indicated by the development of projects such as
the INVENT FVM project (Manstetten et al., 2003) or the TAC Safecar project
(Regan et al., 2001) and more recently by the research activity planned within the
HUMANIST Noe (EC funded network of excellence within the sixth framework
programme). The research planned in the AIDE project will also contribute to this
effort of understanding learning processes and their impact on short-and long-term
behavioural adaptations.
Some interesting concepts have been introduced, such as the concept of 'learn-
ability' and of 'self-explanatory' support systems. With reference to the RE-
SPONSE project (Beker et al., 2001), 'a system is learnable, if accurate assimila-
tion of information by the driver occurs, evidenced in the driver's understanding
of system function, system handling and situational limits'. A 'self-explanatory
ADAS' is defined as a 'driver assistance system leaving a minimal amount of
learning demand to the driver and eliminating learnability issues which can result
in safety-critical traffic situations'.
These concepts are interesting inasmuch as they extend the number of criteria
for assessing the usability of support systems and emphasise the need to take
account of learning issues in the design of driver support systems. These issues
are likely to be particularly crucial in the future development of support systems,
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 157

as the systems will become more complex and integrate multiple driver support
and information functions.

9.4 Behavioural Adaptation in the AIDE Project

As emphasised in the Introduction, most studies of behavioural adaptation have

been short-term studies and the effects of longer practice and experience with
the system are unknown. Furthermore, these studies have mainly dealt with the
use of a single support system. Studying long-term behavioural adaptations and
developing an integrated management of driver support remains a challenge for
research.
This issue is one of the major goals of the AIDE project. In particular, AIDE
aims at generating knowledge and methodologies and developing human-machine
interface technologies for safe and efficient integration into the driving environment
of ADAS and in-vehicle information systems (IVIS), as well as nomad devices
(such as 'personal digital assistants' - PDA - or 'communicators').
Within the AIDE project, a specific research activity is dedicated to deepening
the analysis of short- and long-term behavioural adaptation associated with the use
of various support systems (Saad et al., 2005) and to determining the most relevant
parameters that can be implemented in models for supporting design and safety
assessment processes (Cacciabue and Hollnagel, 2005).
The planning of the research activity with respect to the issue of behavioural
adaptation deals firstly with the problem of the circumstantial and temporal man-
agement of the assistance provided by various systems in the driving process. The
systems studied are ADAS dedicated to the main safety critical driving tasks, that
is, time-headway, speed and lateral control tasks. The systems varied according to
their modes of action (warning or direct intervention) and their degree of adapt-
ability to the situational context or to the drivers' characteristics. Furthermore, the
planned studies provide an opportunity for examining the effects of combining
several ADAS during the driving process.
With respect to the circumstantial conditions that affect processes of behavioural
change, the following aspects are studied:
The nature and extent of behavioural changes associated with the use of various
driver support systems;
The conditions in which these changes take place;
The 'reasons' why these changes occur; and
The characteristics of the drivers more likely to present these behavioural
changes.
This entails in particular (1) gauging the occurrence, nature and magnitude of
the behavioural changes as a function of the type of support system being studied
(informative, prescriptive or intervening system); (2) then examining to what extent
these changes are observed in relation to analogous driving situations or tasks
and/or in relation to drivers' common characteristics.
158 Saad

With respect to the temporal factors affecting behavioural adaptation, two main
phases are considered, namely:

Learning and appropriation phase. During this phase, drivers discover the sys-
tems, learn how they operate, identify the precise limits of their competence and
delimit their domains of utility. This learning phase is assumed to be crucial for
drivers to build an appropriate model of the operation of the systems and for
'calibrating' their trust in them. It is also assumed that the learning process is
oriented by the way the systems are presented to the driver (instruction for use
in the manuals, for instance) as well as by the information and feedbacks they
received on-line when interacting with them.
Integration phase. It is assumed that, through experience gained with the systems
in various driving situations, the drivers are able to assess whether or not, how and
in which situational context, it is possible to integrate them in the management
of the overall driving task.

This involves examining whether, when and how behaviour associated with the use
of support systems changes with training and experience. It has to be pointed out
that, because of the scarcity of research carried out into the learning process and the
long -term effects, it is hard to determine the temporal span of the different phases
distinguished above. Different support systems will probably require different
learning and integration times. The research planned in the AIDE project has been
developed in such a way as to optimise the opportunities for identifying the main
'stabilisation phases' of the learning and integration process for different support
systems.
This research activity will lead to the identification of the relevant variables
to be used to assess behavioural adaptation effects. The correlation between these
variables and adequate taxonomies and classification of road situations and driving
tasks (scenarios of dynamic situation) will be devised to associate the variables
with realistic conditions. On this basis, it will be possible to develop a model of
driver behaviour that can act as reference for the design of an adaptive-integrated
in vehicle interface supporting the multiple tasks of drivers in modem vehicles
(Cacciabue and Hollnagel, 2005).

References
Allen, T., Lunenfeld, H. and Alexander, G. (1971). Driver information needs. Highway
Research Record, 366, 102-115.
Almqvist, S. and Nygard, M. (1997). Dynamic speed adaptation: A field trial with automatic
speed adaptation in an urban area. Bulletin 154. Lund Institute of Technology, Department
of Traffic Planning and Engineering, University of Lund, Sweden.
Amalberti, R. (1996). La conduite des systemes if risques. Presses Universitaires de France,
Paris.
Becker, S., Johanning, T., Feldges, 1. and Kopf, M. (2001). RESPONSE. The integrated
approach of user, system, and legal perspective. Final report on recommendations for
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 159

testing and market introduction of ADAS. Commission of the European Communities,

DG XIII, Project TR4022, Deliverable No. D2.2.
Brown, J.D. (1985). Concepts and definitions in road safety. In M.B. Biecheler, C. Lacombe
and M. Mulhrad (Eds.). Evaluation 85. Proceedings of the International Meeting on the
Evaluation of Local Traffic Safety Measures. Paris, France, pp. 413-422.
Cacciabue, P.C. and Hollnagel, E. (2005). Mental model of drivers: A review of criteria,
variables, and parameters. In L. Macchi, C. Re and P.C. Cacciabue (Eds.). Modelling
driver behaviour in automotive environments. Proceedings of HUMANIST Workshop.
Ispra, Italy, pp. 185-196.
Carsten, O. and Fowkes, M. (2000). External vehicle speed control. Executive summary of
project results. University of Leeds, Leeds, UK.
Comte, S. (2000). New systems: New behaviour? Transport Research, Part F, 3, 95-111.
Evans, L. (1985). Human behavior feedback and traffic safety. Human Factors, 27(5), 555-
576.
Fairclough, S.H., May, A.I. and Carter, C. (1997). The effect of time headway feedback on
following behaviour. Accident Analysis and Prevention, 29(3),387-397.
Fancher, F., Ervin, R., Sayer, 1., Hagan, M., Bogard, S., Bareket, Z., Mefford, M. and
Haugen, J. (1998). Intelligent cruise control field operation test. Final Report. NHTSA
Report No. DOT HS 808849.
French, D.I., West, R.I., Elander, 1. and Wilding, J.M. (1993). Decision-making, driving
style, and self reported involvement in road traffic accidents. Ergonomics, 36(6), 627-
644.
Grayson, G.B. (1996). Behavioural adaptation: A review of the literature. TRL Report 254,
Transport Research Laboratory, Crowthorne, England.
Hjalmdahl, M. and Varhelyi, A. (2004). Effects of an active accelerator pedal on driver
behaviour and traffic safety after long-term use in urban areas. Accident Analysis and
Prevention, 36, 729-737.
Hoedemaeker, M. and Brookhuis, K. (1998). Behavioural adaptation to driving with an
adaptive cruise control (ACC). Transport Research, Part F, 1, 95-106.
Hoedemaeker, M and Kopf, M. (2001). Visual sampling behaviour when driving with
adaptive cruise control. In Proceedings of the 9th International Conference on Vision
in Vehicles. Australia.
Hogema, J.H. and Janssen, W.H. (1996). Effects of intelligent cruise control on driving
behaviour: A simulator study. TNO report TM-96-COI2. TNO Human Factors Research
Institute, Soesterberg, The Netherlands.
Jonah, B.A., Thiessen, R. and Au-Yeung, E. (2001). Sensation-seeking, risky driving and
behavioral adaptation. Accident Analysis and Prevention, 33, 679-684.
Kopf, M. and Nirschl, G. (1997). Driver-vehicle interaction while driving with ACC in
borderline situations. In Proceedings of the 4th World Congress on Intelligent Transport
Systems. Berlin, Germany.
Kopf, M. and Simon, 1. (2001). A concept for a learn-adaptive advanced driver assis-
tance system. In Proceedings of the Conference on Cognitive Science Approaches 2001.
Neubiberg.
Malaterre, G. and Saad, F. (1986). Les aides ala conduite: definitions et evaluation. Exemple
du radar anti-collision. Le Travail Humain, 49(4),333-346.
Manstetten, D., Krautter, W., Engeln, A., Zahn, P., Simon, 1., Kuhn, F., Frank, P., Junge,
M., Lehrach, K. and Buld, S. (2003). Learnability of driver assistance systems - Invent
FVM- driver behavior and human machine interaction. In Proceedings ofthe 10th World
Congress on Intelligent Transport systems. Madrid, Spain.
160 Saad

Michon, lA. (1993). Generic intelligent driver support, a comprehensive report on GIDS.
Taylor and Francis, London.
Muir, B.M. (1987). Trust between humans and machines, and the design of decision aids.
International Journal ofAviation Psychology, 8, 47-63.
Nilsson, L. (1995). Safety effects of adaptive cruise controls in critical traffic situations.
In Proceedings of the 2nd World Congress on Intelligent Transport Systems. Yokohama,
Japan,pp.1254-1259.
Nilsson, L., Harms, L. and Peters, B. (2002). The effect of road transport telematics. In
P Barjonnet (Ed.). Traffic psychology today. Kluwer Academic Publishers, Boston, pp.
265-285.
Noy, I. (1997). Ergonomics and safety of intelligent driver interfaces. Lawrence Erlbaum
Associates, Mahwah.
OECD (1990). Behavioural adaptations to changes in the road transport system. Organi-
zation for Economic Co-operation and Development, Paris.
Parkes, A.M. and Franzen, S.F. (1993). Drivingfuture vehicles. Taylor and Francis, London.
Regan, M.A., Mitsopoulos, E., Tomasevic, N., Healy, D., Connelly, K. and Williams, L.
(2001). Behavioural adaptation to in-car ITS technologies: Update on the Australian
TAC safecar project. In Proceedings of the 8th World Congress on Intelligent Transport
Systems. Sydney, Australia.
Persson, H., Towliat, M., Almqvist, S., Risser, R. and Magdeburg (1993). Speed limiters for
cars. In A field study of driving speeds, driver behaviour, traffic conflicts and comments
drivers in town and city traffic. Report of Department of Traffic Planning and Engineering,
University of Lund, Sweden.
Rudin-Brown, C.M. and Parker, H.A. (2004). Behavioural adaptation to adaptive cruise
control (ACC): Implications for preventive strategies. Transport Research, Part F, 7,
59-76.
Saad, F. and Malaterre, G. (1982). Regulation de la vitesse: aide au controle de la vitesse.
Synthese. Rapport ONSER.
Saad, F. and Villame, T. (1996). Assessing new driving support systems - Contribution
of an analysis of drivers' activity in real situations. In Proceedings of the Third Annual
World Congress on Intelligent Transport Systems. Orlando, USA.
Saad, F. and Villame, T. (1999). Integration d'un nouveau systeme d'assistance dans
l'activite des conducteurs d'automobile. In l-G. Ganascia (Ed.). Securite et cognition.
Editions Hermes. Paris, pp. 105-114.
Saad, F., Munduteguy, C. and Darses, F. (1999). Managing interactions between car drivers:
An essential dimension of reliable driving. In Proceedings of Seventh European Con-
ference on Cognitive Science Approaches to Process Control (CSAPC'99). Villeneuve
d' Ascq, France. Presses Universitaires de Valenciennes, pp. 99-104.
Saad, F. (2002). Ergonomics of the driver's interface with the road environment - The
contribution of psychological research. In R. Fuller, and lA. Santos, (Eds.). Human
factors for highway engineers. Pergamon, Oxford, pp. 23--41.
Saad, F., Hjalmdahl, M., Canas, r, Alonso, M., Garayo, P., Macchi, L., Nathan, F., Ojeda,
L., Papakostopoulos, V., Panou, M. and Bekiaris, A. (2004). Literature review -Analysis
of behavioural changes induced by ADAS and IVIS. AIDE Project, Deliverable D1_2_1.
Saad, F., Bekiaris, A., Brouwer, R., Carsten, 0., Hjalmdahl, Hoedemaker, M., Ojeda, L.,
Papakostopoulos, V., Nathan, F. and Vezier, B.(2005). General experimental plan for
short and long term behavioural assessment. AIDE project IST-I-507674-IP, Deliverable
Dl_2_2.
 9. Dealing with Behavioural Adaptations to Advanced Driver Support Systems 161

Smiley, A. (2000). Behavioural adaptation, safety and intelligent transportation systems.

Transport Research Record, 1724, 47-51.
Stanton, N.A., Young, M. and McCaulder, B. (1997). Drive-by-wire: The case of driver
workload and reclaiming control with adaptive cruise control. Safety Science, 27(2/3),
149-159.
Suchman, L.A. (1987). Plans and situated actions. The problem of human-machine com-
munication. Cambridge University Press, Cambridge.
Tornros, 1., Nilsson, L., Ostlund, J. and Kircher, A. (2002). Effects of ACC on driver be-
haviour, workload and acceptance in relation to minimum time headway. In Proceedings
of the 9th World Congress on Intelligent Transport Systems. Chicago, IL, USA.
Van der Laan, J.D., Heino, A. and De Waard, D. (1997). A simple procedure for the assess-
ment of acceptance of advanced transport telematics. Transport Research, Part C, 5( 1),
1-10.
Varhelyi, A. and Makinen, T. (2001). The effects of in-car speed limiters: Field studies.
Transport Research, Part C, 9(3), 191-211.
Ward, N.J., Fairclough, S. and Humphreys, M. (1995). The effect of task automatisation in
the automotive context: A field study of an autonomous intelligent cruise control system.
In D.J. Garland and M.R. Endsley (Eds.). Experimental analysis and measurements of
situational awareness. Embry-Riddle Aeronautical University Press, Daytona Beach, pp.
369-374.
Weinberger, M., Winner, H. and Bubb, H. (2001). Adaptive cruise control field operational
test - The learning phase. JSAE Review, 22, 487--494.
Young, M. and Stanton, N.A. (1997). Automotive automation: Investigating the impact on
drivers' mental workload. International Journal of Cognitive Ergonomics, 1,325-336.
Young, M.S. and Stanton, N.A. (2004). Taking the load off: Investigations of how adaptive
cruise control affects mental workload. Ergonomics, 47(9), 1014-1035.
 IV
Modelling Motivation and
Psychological Mechanisms
10
Motivational Determinants of Control
in the Driving Task
RAY FULLER

10.1 Introduction
Road transport is mainly about moving people and goods in motorised vehicles.
Since the introduction of such vehicles in the nineteenth century, people have
had the task of controlling them, although developments in technology today are
moving rapidly towards the possibility of displacing the human element. For the
time being, however, a human driver is in control and is faced at each moment
on a journey with two fundamental choices, the direction in which to steer the
vehicle (i.e., path choice) and the speed at which to move. In trying to understand
how drivers make these choices, this paper will focus primarily on the choice of
speed. Choice of speed is not only a much less constrained choice than choice of
path (except in congested traffic) but it is choice of an inappropriate speed which
contributes significantly to a very large proportion of collisions and road run-offs.
As has frequently been written and perhaps all too frequently forgotten, driving is
a self-paced task and it is the freedom of this self-pacing that underlies so much
of what can go wrong. Despite the focus on speed choice here, however, choice
of path or direction is not entirely neglected: The emerging understanding of the
fundamental determinant of speed choice will be seen to apply equally well to
choice of direction.

10.2 Understanding Speed Choice

10.2.1 Behaviour Analysis
So how might we begin to understand drivers' choices about speed? One explana-
tory framework that is potentially useful as a starting point is that of behaviour
analysis and the fundamental processes of operant conditioning. In its simplest
form, this theoretical approach eschews consideration of what might be going
on in the driver's head (except, perhaps, observable parallel processes of neural
activity) and confines itself to describing systematic relationships amongst observ-
able events. The basic paradigm of this approach is represented by a three-term

165
166 Fuller

contingency of stimulus condition (or more specifically, discriminative stimulus),

the behaviour of interest and the consequences of the behaviour.
Under particular stimulus conditions, behaviour that is followed by reward-
ing consequences becomes reinforced or strengthened relative to alternative be-
haviours which could occur under those conditions. Thus, in a particular road
environment (stimulus condition), the speed 'choice' (behaviour) that is followed
by rewarding consequences (e.g., maintenance of control of vehicle while contin-
uing to attain travel goals) becomes strengthened relative to other speed 'choices'.
Over time, particular stimulus conditions (discriminative stimuli) come to signal
this behaviour--eonsequences relationship. These contingencies can then be de-
scribed as an implicit internalised rule which tells the driver what to do under
particular conditions (e.g., 'if it's raining, drive more slowly'). Through this pro-
cess, particular speed 'choices' are gradually learned, initially through a kind of
'trial-and-error' stage (in which the error might be experienced as unpleasantly
high g forces, the beginnings of a skid, a near miss or even a complete loss of con-
trol). Eventually particular speeds for particular conditions become established,
habitual behaviours. Through the further process of stimulus generalisation, what
is learned in one specific context, such as a speed that is rewarding on a bend of a
particular radius, transfers to other similar contexts, without the need for additional
trial-and-error learning and conditioning.
This way of thinking about drivers' speed 'choices' was earlier articulated by
the author in the threat-avoidance model (Fuller, 1984), which focused on the
rewarding consequence of the avoidance of collision. This emphasis was preferred
for two reasons. First, there is the inescapable observation that as soon as a driver
sets the vehicle in motion the probability of collision is one - unless, of course,
frequent actions are taken to avoid this outcome. Thus driving may be considered
to be, in the main, an avoidance task. The second reason was that there was a
reasonably extensive empirical literature on the conditions under which avoidance
responses might be delayed and it was possible in principle to translate the extent
of delay of an avoidance response into a level of statistical risk of collision. Thus
the opportunity presented itself to build on previous work on avoidance learning in
various contexts (admittedly mainly undertaken with species other than humans) in
attempting to understand drivers' speed behaviour. However, this approach suffered
from its very emphasis, namely avoidance learning, and ignored motivation to make
progress. This shortcoming was subsequently dealt with by recasting the analysis
within the broader framework of instrumental learning which encompassed both
types of rewarding consequence: Avoidance of threat and making progress (see
Fuller, 1991a, 1991b).
In parallel with this development and to a large extent converging with it, was
the seminal work of Naatanen and Summala (1976), subsequently developed by
Summala (1986, 1997) and which is sometimes referred to as the zero-risk model.
Summala argues that for most situations drivers have learned what they should
or should not do to avoid a certain or almost certain collision. Driver behaviour,
including speed choice, is determined by the maintenance of safety margins. These
are learned through experience and so most of driving becomes a 'habitual activity
 10. Motivational Determinants of Control in the Driving Task 167

which is based on largely automatised control of safety margins in partial tasks'

(Summala, 1986, p. 10). Thus this model may be seen to be firmly situated in the
well-established instrumental learning paradigm, as described above. The model
further argues that what undermines the maintenance of safety margins, however,
are motivating conditions which push drivers to higher speeds, insensitivity to low
probability events on the roadway and desensitisation to potential threats because
these are not realised.
Although driver behaviour research which makes explicit use of a behaviour
analysis paradigm is not frequently published, there are some notable exceptions
(see, e.g., the work of Geller, 1998; Ludvig and Geller, 2000; Reinhardt-Rutland,
2001; Hutton et al., 2001; Harrison, 2005). Furthermore, it is rather difficult to
escape the paradigm if one wishes to understand from a psychological perspec-
tive the effects of enforcement on driver behaviour, including effects on speed
choice. The implementation of enforcement in relation to traffic regulations, such
as a speed limit, changes the consequences of not complying with the regulations
by making the consequences more punishing. Regulation-compliant behaviour is
thus strengthened to avoid the additional punishment (Fuller and Farrell, 2001).
The effectiveness of this strategy when implemented with sufficient intensity is
testament to the effectiveness of manipulating consequences to achieve changes
in behaviour (Makinen et al., 1999).
There is clear merit in the parsimony of the conceptual framework of be-
haviour analysis and it successfully avoids reliance on hypothetical constructs
(see MacCorquodale and Meehl, 1948). Nevertheless, using the driver's condi-
tioning history of learned relations amongst discriminative stimuli, responses and
consequences to explain the driver's choice of speed are not without serious prob-
lems. This strict behavioural approach is vulnerable to the somewhat implausible
requirement that drivers learn through a prolonged conditioning process how to
respond safely to what is virtually an infinite number of road and traffic scenarios.
There is a huge burden on the process of learning to explain sustained mobility at
a particular speed while avoiding collision. There is no question that experience
contributes significantly to the development of driver competence but, as has been
argued before (Fuller, 2002), the learning of contingencies in the road and traf-
fic system is challenged by a low frequency of opportunities for learning about
infrequent hazards and by uncertain relationships amongst the events being ex-
perienced. Ranney (1994) has also criticised the behaviour-analytic approach for
its theoretical difficulty in handling embedded contingencies. As Michon has so
succinctly put it (Michon, 1989), there is no place in the model for 'meanwhile'.

10.2.2 The Theory of Planned Behaviour

The cognitive revolution beginning in the early 1960s created a new climate for
the study of internal mental processes as causes of overt behaviour, expressed in
its simplest form as 'what we think determines what we do' (Bern and de Jong,
2006). An example of this kind of approach is the theory of planned behaviour
(Ajzen, 1991). Although it is not usually explicitly recognised, this theory forms
168 Fuller

a clear link with the behavioural approach discussed above, because its implicit
fundamental feature is an internalisation of the behavioural paradigm as a set of
beliefs about each specific behaviour that might be emitted.
In the theory of planned behaviour (TPB), any specific behaviour is determined
by intention. This term does not appear to mean intention in the conventional
sense (i.e., a conscious representation of a plan or purpose) but rather is assumed
'to capture the motivational factors that influence a behaviour; they (intentions)
are indications of how hard people are willing to try, of how much effort they are
planning to exert, in order to perform the behavior' (Ajzen, 1991, p. 181). Thus
the stronger the intention, the more likely the behaviour (assuming it is under
volitional control). Intention is in tum determined by three other constructs or
variables, respectively labelled 'attitude toward the behaviour', 'subjective norm'
and 'perceived behavioural control' (see Fig. 10.1).
The variable 'attitude toward the behaviour' is influenced by beliefs which link
the behaviour to certain outcomes or some other attribute such as the cost incurred
in performing the behaviour (both are types of consequence). Thus beliefs about
behaviours internalise consequences as important determinants of what behaviour
is learned: 'We learn to favour behaviours we believe have largely desirable conse-
quences and we form unfavourable attitudes toward behaviours we associate with
mostly undesirable consequences (op. cit. p. 191).'
The variable' subjective norm' represents a specific kind of social consequence,
expressed as perceived social pressure to perform or not to perform the behaviour
in question. This social pressure is determined by the degree of approval or dis-
approval (i.e., rewarding or punishing social consequences) of the behaviour by
important referent individuals or groups.
Finally in Ajzen's theory, 'perceived behavioural control' is based on beliefs
about the availability of opportunities and resources for performing the behaviour
of interest. This aspect of the belief structure in the theory may be regarded as an
internal representation of the discriminative stimulus which, as described earlier,
plays a key role in behaviour analysis - specifying the conditions which signal
the behaviour-consequences contingency. Interestingly in behaviour analysis it is
argued that over time a discriminative stimulus can come to exert direct control
over behaviour, the concept of stimulus control. In the same way, this notion is
reflected in TPB by the assertion that there can be a direct determining link between
'perceived behavioural control' and the behaviour itself (see Fig. 10.1).
So the step in the direction of a cognitive explanation characterised by TPB
essentially provides for a mental representation of the contingencies of the be-
haviour analysis paradigm, but with an added emphasis on social consequences of
behaviour. How does it fare? According to Ajzen (1991), the theory is designed to
predict human behaviour in specific contexts. If the theory successfully includes all
of the determinants of behaviour, then all variance in behavioural outcome should
be predicted, except for any residual arising out of measurement error. Each in-
dividual behaviour should have associated with it a particular pattern showing
the relative contribution of each variable in the theory in the determination of
that particular behaviour and it may be noted that in the individual case it is not
 10. Motivational Determinants of Control in the Driving Task 169

attitude
toward the
behaviour

subjective
intention behaviour
norm

perceived
behavioural
control

FIGURE 10.1. A simplified representation of the constructs of the theory of planned be-
haviour (arrows identify potential determining relationships).

necessary that all variables should have an equal or indeed any influence. In a
relatively recent meta-analysis performed by Armitage and Connor (2001), it was
found that the TPB model accounted for only about 27% of the variance in actual
behaviour. Furthermore, if reinforcement history is included as an additional pre-
dictor, operationalised as frequency of the behaviour in the past, the increase in
explained variance can be anything from 5% to 32% (Ajzen, 1991). Thus, although
one published study has demonstrated that the TPB can provide some account of
actual speeding behaviour (Elliott et al., 2003), the outlook for the theory in this
respect does not look very promising. Perhaps its real strength for the moment is
in its potential for separately identifying key beliefs which are strongly related to a
particular behaviour, beliefs which might then provide the focus for interventions
to change that behaviour.

10.2.3 Risk Homeostasis Theory

An alternative cognition-situated solution to the question of what determines a
driver's choice of speed is the suggestion that drivers carry out a risk evaluation
of speed (and other) response alternatives and settle on the level of risk that op-
timises net benefit. This approach is represented in its most developed form in
Risk Homeostasis theory and the concept of Target risk (e.g., Wilde, 2001), but a
similar utility model is also entertained by other researchers such as Janssen and
Tenkink (1988) and Deery (1999). A simplified version of the key components of
Wilde's model is presented in Fig. 10.2.
In Wilde's model, drivers weigh up the costs and benefits of alternative actions
and this results in an accepted level of risk which they actively target (target risk).
On a continuous basis, this target level of risk is compared with the perceived
level of risk arising from the driver's actions in relation to the road and traffic
environment. Drivers then adjust their behaviour as discrepancies between target
and perceived risk arise. The decision-making process is thus characterised as a
170 Fuller

benefits and target risk decision and

costs of safe response
and unsafe
behaviour

perceived risk vehicle speed

and direction

FIGURE 10.2. A simplified representation of key functional elements of risk homeostasis

theory.

closed-loop homeostatic system. Thus if a driver perceives that current speed on

a particular bend is at a risk level that exceeds the target risk level, then the driver
will slow down until the perceived and target levels converge. In the same manner,
if current speed is perceived to be at a risk level lower than target risk, then the
driver will speed up until, again, perceived and target levels converge (it may be
noted that this example simplifies drivers' options somewhat - alternative ways
of modifying risk may also be employed, such as engaging in a secondary task -
however the speed manipulation is clearly the dominant one in typical driving).
Wilde argues that the aggregation of individuals' target risk levels produces the
accident toll in the drivers' jurisdiction over a period of time. Thus target risk and
accident rates covary. An important implication of this conceptualisation is that
safety interventions such as safer vehicle design, improved roadway design and
improved driver training and assessment are all doomed to failure. Their safety
impact is by and large traded for increased speed. The only way to reduce accidents
on the road (and in general), therefore, is to reduce the target level of risk aimed
for by drivers.
However the validity of this conceptualisation founders on certain kinds of ev-
idence. The probabilities associated with collisions on the roadway are extremely
low, with the average risk for a US driver, for example, being approximately equiv-
alent to one crash every 5 years (Evans, 1991). If such a risk level is distributed
over each and every decision made by a driver over that length of driving period,
the distributed risk estimation emerges as being completely beyond human com-
putational capability (Slovic et al., 1977). Furthermore, drivers show incomplete
compensation for various safety interventions. For example, Rumar et al. (1976)
found that drivers with studded tyres on snow compared with drivers without such
tyres did indeed drive faster. However, they did not drive so much faster that they
cancelled out the added safety benefit of the studded tyres; there remained a signif-
icant decrease in risk of loss of traction. Similarly, as published in an DEeD review
of evidence for adaptation by drivers to safety measures, increases in lane width
cause increases in average speed (1 to 2 mph per foot) - but there is also a reduc-
tion in accidents; the addition of a paved shoulder to two-lane rural roads increases
speeds by up to 10% - but decreases accidents by up to 40%, and the addition of
 10 . Motivational Determinants of Controlin the DrivingTask 171

edge-lines to two-lane rural roads increases average speeds, but decreases accident
frequency and severity (OECD, 1990). The report concluded that although there
was some evidence of risk compensation under certain conditions - there was little
to support the concept of risk homeostasis. Drivers are clearly able to adapt their
behaviour in varying ways, to varying extents and to varying conditions. However
until relatively recently, there has been no account to explain the conditions under
which behavioural adaptation may occur. An attempt to do this and to describe the
components that influence driver decision making in a comprehensive way is the
task-capability interface model (Fuller, 2000; Fuller and Santos, 2002) .

10.2.4 The Task-Capability Interface Model

The task-capability interface (TCI) model starts with the self-evident truth that a
loss of control by the driver necessarily arises when the demands of the driving
task exceed the available capability of the driver and that control is maintained
when those task demands are less than the driver's available capability. When loss
of control occurs, this may lead to a collision (or road run-off) . On some occasions,
however, the driver may be able to regain control without further mishap or the
task may abruptly change by a potential collision-object changing course at the last
moment, thereby getting out of the way of the approaching vehicle (see Fig . 10.3).
One implication of this is that it makes more sense to refer to loss of control of the
driving task as the key marker of safety in driving. There has been a consensual
move to drop reference to the tenn 'accident' as the key marker, with collision

,,r-- - - - - ---- - -- -- - - - -- -''

I I

: LUC KY ESCA PE :
,
I
'
'
I '
- -- --.-- - -- --- - - - ~ -_.
, I
, I
I r--------'------- :- - - - - - - - - - - -
: I :
: : co mpe nsatory: : CO LLIS ION
I : ac tion by others ' .. ' !
:I ~ ---- -- - -r---
I
--""-~ ~ ; ; :1- - - - - - - - - - - -
.------1- -- --. t.. . . . .... ..

I
I I .......

: LOSS OF :, /
C A PA BI LI T Y (C) : CONT ROL:
~---------------------
<. .>
~ ......
>"
:_ -- ---------~
C<D
C>D
, ,k """ T ASK
,, DEM AND S (D )
,,
,,: CONTRO L ,:
1
, --- - -- -
,,

FIG URE 10.3. The basic conditions for control of the driving task.
172 Fuller

being the current preferred term. In principle, however, although collisions are
frequently the outcome of failed driver performance and are the significant events
from a safety perspective, they are nevertheless not a necessary outcome.
We can explore the Tel model further by unpacking the elements of driver ca-
pability on the one hand and driving task demands on the other. Driver capability is
bounded by the constraints imposed by the biological characteristics of the driver,
constraints associated with for example the effectiveness and efficiency of sensory
and perceptual processes, information processing capacity and speed, speed of mo-
tor response, motor coordination, flexibility, strength and physical reach. Starting
with these constitutional biological characteristics, education, training and experi-
ence each contribute to the development of knowledge and skills. Such knowledge
includes formal elements such as rules of the road, procedural knowledge defining
what to do under what circumstances (conditional rules) and a representation of
the dynamics of road and traffic scenarios which enable prediction of how those
scenarios will develop (like an internalised mental video which runs on ahead of the
immediately-observed situation). Skills include basic vehicle control skills as well
as handling skills in challenging circumstances (such as skidding). Together these
biological characteristics and acquired characteristics through education, training
and experience determine the upper limit of competence of the driver. However,
this competence is not necessarily available at every moment. Performance is vul-
nerable to a host of variables which include motivation, fatigue, drowsiness, time
of day, drugs, distraction, emotion (such as fear, anger and aggression), stress and
level of effort. Any of these can undermine driver competence to yield a level of
capability at a somewhat lower level (see Fig. 10.4). We might label these variables
collectively as human factor variables. A further set of human factor variables re-
late to motivation for speed and therefore have an effect not so much on driver
capability but rather on the demand level of the driving task.
Driving task demands are determined by a range of interacting elements (see
Fig. 10.4). These include first the physical environmental factors such as visibility,
road alignment, road marking, road signs and signals, road surfaces and curve
camber angles and so on. Secondly, there are other road users with various prop-
erties including that of occupying or the imminent potential of occupying, critical
space in the projected path of the driver. Thirdly, there are the operational features
of the vehicle being driven, such as its information display characteristics, control
characteristics of steering, braking and accelerating and its capability to provide
roadway illumination in dark conditions. Finally and perhaps most important of
all, there are elements of task difficulty over which the driver has immediate and
direct control, namely the vehicle's trajectory and speed. Of these speed is clearly
the most significant factor: It is self-evident that the faster a driver travels, the
less time is available to take information in, process it and respond to it and the
less time there is to correct any emergent error. As mentioned earlier, the driving
task is a self-paced task and this means that in the last analysis driving task de-
mand is under the control of the driver (exceptions to this rule are where a driver
complies with a speed limit which is slower than the driver's preferred speed;
where a driver is under pressure to comply with a schedule or where a minimum
 10. Motivational Determinants of Control in the Driving Task 173

biological
characteristics

ed ucation/tra ining!
experience ,,r- - - - - - - - - - - - - - -- - --- -''
, '
: LUC KY ESCA PE :
,, '
...
'- - - - - - - - - - - --- - - :.: - - ~
'

,
,
,I

: i- -- - - - - - '- -- --- j :- - - - - - - - - - - -
i : co.mpe nsatory i : CO LLIS ION
: i ac no n by others : ~.. : !
, 1 -_-_-,----- -[~~~-::.::;'/ :------------

I CA PA BILITY (C)
L - -_

"" "" - , ' ..

i
i
'
1

...............
LOSS OF
CONT RO L

--
:--/~
i
_,

......y ....

other road user s

t
11'---------
vehicle properti es
I

r human factor variables I

FIGURE 10.4. Elements of the task-capability interface model.

speed is required over a road section). Each of these task demand variables may
independently contribute to the level of task demand and they may also interact in
generating that demand. Furthermore, human factor variables may influence speed
choice, which in tum may influence other human factor variables such as arousal
level (see Fuller, 200Sa). These determinants of task demand and the determinants
of driver capability described above are brought together in the representation of
the task-capability interface model presented in Fig . lOA.
Thus far, what we have is a descriptive model of classes of variable which interact
at the interface between capability and task demand to determine the outcome for
the driver in terms of maintenance of control. We can now conceptualise and define
the difficulty of the driving task in terms of the degree of separation of task demand
and capability, with high difficulty where there is little separation and low difficulty
where there is large separation. Difficulty level as here defined is proportional to
the inverse of spare capacity, with spare capacity shrinking as difficulty level rises
174 Fuller

biological characteristics
education, training
experience

I
human factor variables
t
~ effort

1--------------
I

I
CA PA BILITY

inver se of TAS K DI FFI C ULT Y (or wor kloa d)

(spare capacity , safety margin)

1-------------- TAS K DEM AND

t t t
vehicle, environment
road users. speed

FIGURE 10.5. The determinants of task difficulty.

(see Fig. 10.5). It may also be considered to be equivalent to the concept of mental
workload (de Waard, 2002), although a more comprehensive equivalence would
need to include physical workload as well.
As a model, the TCI model provides a conceptual framework for organizing
the critical variables which generate potential hazard scenarios and the conditions
under which the driver will lose control of the driving task and become vulnerable
to the range of possible disastrous consequences which might ensue . At this level of
description the model is mainly behavioural in the sense that it is largely confined
to describing observable phenomena (driver performance characteristics, history
of learning, training and experience and human factor variables on one side of
the key interface between capability and task demand and the characteristics of
the road and road-user environment, the vehicle, vehicle trajectory and speed
on the other). Note that this approach avoids a difficulty which is not addressed in
some other formulations (e.g., Deery, 1999) namely the important question of how
to define a hazard : This is determined in the model as the outcome of the interface
or transaction between capability and task demand.

10.2.4.1 The Determination of Task-Difficulty Level :

Task-Difficulty Homeostasis
Thus far, recour se to a cognitive level of description has not been invoked. However
as it stands, the TCI model simply provides a snapshot of key interacting elements
 10. Motivational Determinants of Control in the Driving Task 175

perceived
capability; range of
journey goals; acceptable
effort task difficulty
motivation;
characteristic
difficulty level
preference decision and
perceived response
r----- task difficulty
I
I
I
I
I
I objective task
~
I difficulty effects on vehicle speed
I
I and position
I
~------------------------------------------ and
on other road users

FIGURE 10.6. Representation of the process of task-difficulty homeostasis.

in the complex dynamic of a driver engaging with the road and traffic system. One
important question then arises from a safety perspective and that is the question
of what determines the level of task difficulty that pertains at any moment of time.
It is self-evident that this is clearly not a random state that emerges quite inde-
pendently of driver determination. As stated earlier, driving is a self-paced task
and by manipulating the speed of the vehicle the driver has direct control over the
most important single determinant of task demand. The Tel model thus specifies
one key hypothesis, namely that drivers drive in such a way as to maintain level
of driving task difficulty within a preferred range. This hypothesis amounts to one
of task-difficulty (or workload) homeostasis, as represented in Fig. 10.6. Based
on the goals of a particular journey, self-appraisal of capability, effort motivation
and perhaps a reasonably stable preferred level of difficulty characteristic of the
individual, a driver' selects' a range of difficulty within which she or he is prepared
to operate and drives in such a way as to maintain experienced difficulty within
that range. Manipulation of speed is the primary mechanism for achieving this,
although undertaking or dumping other tasks secondary to the primary driving task
may be used on occasion (see, e.g., Hart and Wickens, 1990). This hypothesis of
task-difficulty homeostasis provides an intriguing explanation for why inexperi-
enced drivers are so vulnerable to loss of control and collision: They are liable to
underestimate task difficulty by (a) overestimating capability on the one hand and
(b) underestimating task demand on the other (some relevant evidence is reviewed
in Fuller, 2000). The hypothesis also provides a theoretical basis for the design of
safety interventions, for example by making the driving task appear more difficult
than it really is in situations where a reduction in driving speed is warranted. How-
ever what is more important for this discussion is to examine next the evidence for
the hypothesis of task-difficulty homeostasis.
176 Fuller

s:
Hpreferred speed i ;.1
Cl
.E ~r
.-
.r
I ---+-
Task difficulty
___ Statistical risk
" ./ ~ j Feelings of risk

f - - -. /
r:
,/,
/
/
10 speed hi

FIGURE 10.7. Stylised representation of driver ratings of task difficulty, statistical risk,
feelings of riskand preferred speedforroadsegments viewed at systematically incremented
speeds.

One strong prediction from the hypothesis is that, if all else is held constant,
task difficulty should correlate with speed. A second prediction is that estimates of
statistical risk should be zero until task demand begins to approach capability, that
is when difficulty level exceeds some criterion value. This point would correspond
with the driver 's perception of a real hazard emerging in the road and traffic
situation. A third prediction is that a driver's preferred level of difficulty should
be at a speed lower than that at which estimates of the statistical risk of loss-of-
control and crashing rise above zero. This last provides a further test of Wilde 's
target risk hypothesis. We have examined these predictions by soliciting drivers'
ratings of task difficulty and preferred speed as they viewed video-tapes of the
same segment of roadway travelled at different speeds . We also got the drivers
to provide an estimate of statistical risk by imagining travelling each segment at
a particular speed on 30 separate occasions and indicating how many times they
thought they would lose control. This technique was devised to enable drivers to
avoid the difficulty associated with providing a probability estimate of a single
rare event, at the same time enabling use of a meaningful estimate (e.g., 2 times
out of 30; 0.1 times out of 30, etc .).
The results confirmed both predictions one and two (see Fig. 10.7). Task dif-
ficulty correlated very highly with speed (speed accounted for over 98% of the
variance in ratings of difficulty) and over a range of slow to moderate speeds,
statistical risk ratings remained at zero and did not begin to rise until speeds were
very much faster. We obtained the same pattern for different types of roadway and
repeated the results in a subsequent replication (see Fuller et aI., 2006, in press).
In relation to the test of Wilde's Target risk hypothesis, as can be seen from the
vertical line in Fig. 10.7, the driver 's preferred speed was lower than the point at
which statistical risk exceeded zero, providing further evidence against the Target
risk concept. However the simulation studies threw up an unexpected discovery.
Along with ratings of task difficulty and ratings of statistical risk , we also asked
drivers to rate their feelings of risk, fully expecting those ratings to track their
statistical risk ratings . There seemed to be no a priori reason why feelings of
risk should be dissociated from ratings of statistical risk and there seemed to
 10. Motivational Determinants of Control in the Driving Task 177

be no reported evidence in the literature to suggest such dissociation. But our

expectations were confounded. Ratings of feelings of risk did not track statistical
risk: What they did track were the ratings of task difficulty (see Fig. 10.7). The
correlation between the two variables was of the order of 0.97 (Fuller et al., 2006,
in press). This finding raises an intriguing question. Why should task difficulty
and feelings of risk be so strongly associated? Is it through a feeling of risk, which
has been shown to be different from an estimate of statistical risk, that drivers
sense task difficulty? Taylor (1964) concluded from a study of the relationships
between driver arousal (operationalised as GSR level), speed and road segment
characteristics that 'drivers adopt a level of anxiety that they wish to experience
when driving and drive so as to maintain it'. Similarly, Naatanen and Summala
(1976) have long advocated the concept of a subjective risk (or more recently
'fear') monitor in driver decision making, although Summala (1986) has argued
that zero-risk experience of road and traffic leads to a desensitisation to emotional
feedback. In the last decade, one researcher who has strongly advocated the role of
feelings in decision making is Damasio (1994, 2003). Vaa (2003) has subsequently
developed his own driver behaviour model predicated on this viewpoint (see also
this volume).

10.2.4.2 The Representation of Task-Difficulty

Before considering the possibility of feelings of risk as the determinant of driver
speed choice, let us take a step back to consider the more general problem. If
the Task-difficulty homeostasis hypothesis is correct, then the question of how
drivers represent and determine task difficulty or workload is fundamental to de-
veloping our understanding further. This question may be re-phrased as 'how is
task difficulty represented in the 'comparator' element of the process depicted in
Fig. 10.6?'
If the analysis of task-difficulty here was restricted to physical workload in
dynamic work, then the answer to this question would be reasonably straight-
forward (see, e.g., Obome, 1995). As physical workload increases, there is an
immediate sense of having to expend additional effort. As this additional effort
is made, the demand for oxygenated blood supply increases and to deliver that
there is an automatic increase in respiration rate and a parallel increase in heart
rate. There is, in addition, an increase in heat energy output and typically a re-
flex response of increased sweating in order to increase the rate of heat loss from
the body. All these changes can be sensed by the person undergoing the physical
task. Thus apart from the sense of effort there is feedback about physiological
changes that are needed to support that effort. As physical workload increases
even further, there comes a point where the work requirement approaches the up-
per limit of muscular output of which the person is capable. In this region the
sense of effort required increases significantly and the emerging fatigue in muscle
tissue is experienced not just as fatigue but as an aversive, even painful, condition,
making it even more difficult to sustain the effort. Attempts to increase or even sus-
tain the workload are self-limiting: The muscles are simply unable to work at the
178 Fuller

required level and the accompanying oxygen debt leaves the individual winded and
with pounding heart. Immediate cessation of the level of activity is then the only
option.
The selection by an individual of a particular level of sustainable workload for
a physical task will be determined by the goals of the task, perceived level of
capability (possibly modified by human factor variables such as level of activation
and level of fatigue), the level of effort the person is prepared to make and possibly
a level of preferred workload that is characteristic of the individual (as represented
in the distinction: She or he is or is not a 'hard worker').
Is there a similar system of sensitivity to mental workload (as one possible
representation of driving task difficulty) that parallels that for physical workload?
There is no question that one can be aware that different cognitive tasks demand
different degrees of cognitive effort (see, e.g., Shugan, 1980; Payne et al., 1993) but
there do not appear to be reliable and, more importantly, detectable physiological
changes which underpin that effort which can provide additional feedback to the
individual in the same manner as increased respiration, heart rate, temperature and
sweating as in the case of muscular effort. Nevertheless, of course, there are rate
and/or capacity limits in the cognitive processes of information uptake, working
memory storage and processing. As mental workload or task difficulty increase,
there comes a point where the cognitive requirement approaches these upper limits.
At this stage, just as with dynamic physical work, the sense of effort required will
increase significantly, there may also be a sense of mental fatigue, the condition
may even feel aversive, but there does not appear to be an equivalent sense of pain
(although this possibility was once amusingly portrayed in a Monty Python sketch
in which the intellectually challenged Mr. Gumby complained 'my brain hurts').
As indicated in Fig. 10.6 and just as in the case for a physical task, the selection
of a sustainable workload for a cognitive task (i.e., task difficulty) will be deter-
mined by the goals of the task, perceived level of capability, the level of effort the
person is prepared to make and again possibly a level of preferred workload that
is characteristic of the individual. But without the physical changes which feed
information back to the individual in physical work, how is a particular level of
work or task difficulty detected and controlled? One possibility relates to the goals
of the driving task.
The immediate goals of the driving task are twofold: To achieve a journey and
to maintain control. Control is the primary element here because without control
the journey cannot be completed. So perhaps we can rephrase the question as
'how is control represented in the system and what information provides feedback
regarding the status of control?'
We might speculate that the representation of being in a state of control includes
the ability to make progress, the ability to make avoidance responses where nec-
essary (e.g., of objects or other road users), the maintenance of adhesion to the
road surface, having access to critical information, such as the requirement for
an avoidance response and a rate of flow of information and rate of response re-
quirement that are within the individual's capability. All of these seem self-evident
because control would be lost if any of them were absent. Imagine a vehicle with
 10. Motivational Determinants of Control in the Driving Task 179

no accelerator, no brake, no steering and no forward vision or a vehicle with all of

these but travelling at such a speed that the vehicle had no directional control or
the rate of flow of information exceeded information processing limits or critical
information arrived so late that there was no time opportunity to respond to it. Now
given that the ability to accelerate, brake, steer and have forward vision is rela-
tively stable characteristics of driving, the critical variables representing control
become maintenance of contact with the road surface, rate of flow of information
and available response time.
How might a human metacognitive system monitor adhesion, rate of flow of
information and available response time in order to determine whether or not they
are within acceptable limits? Is there a TOTE (test-operate-test-exit) mechanism
(Miller et al., 1960) evaluating the status of each variable through a continuous,
recursive control loop, against a set of tolerance criteria, triggering adjustment
responses before the next test and so on? Is there a mechanism that recognises a
decoupling when, say, information arrives at such a rate or arrives so late that it
cannot drive output (see, e.g., the discussion of augmented cognition by Young
et al., 2004)?
In the context of driving, one further possibility is that if the value of any of
these variables exceeds the upper limits of tolerance it may trigger feelings of an
impending loss of control and, given the potential punishing consequences of this,
anxiety and fear. Such a process would clearly have an adaptive function in terms
of motivating the individual to keep within safe boundaries of operation. Thus is
there some critical loss of adhesion (and therefore directional control) or critical
rate of information flow or response requirement which triggers feelings of anxiety
and fear? Our evidence of the close association between feelings of risk and level
of task difficulty is certainly highly consistent with such a suggestion and, as
mentioned earlier, one researcher who has strongly advocated the role of feelings
in decision making is Damasio (1994, 2003).

10.2.5 The Somatic-Marker Hypothesis

Although normal language usage regards emotion as a type of feeling, Damasio
(2003) reserves the term 'emotion' for underlying body states (perhaps triggered
by some precipitating event) and the term' feeling' for the experience of these states
(rather akin to the James-Lange theory of emotion). He concludes that emotions
provide a natural means for the brain and mind 'to evaluate the environment within
and around the organism and respond accordingly and adaptively' (p. 54). Whether
or not one is paying attention, Damasio suggests that emotionally competent stimuli
(i.e., stimuli with which some feeling is associated) can be detected and that
attention and thought can then be diverted to those stimuli, thereby enhancing the
quality of reasoning and decision making (an orienting role for emotion, earlier
suggested by Zajonc, 1980). Emotional signals 'mark options and outcomes with
a positive or negative signal that narrows the decision-space and increases the
probability that the action will conform to past experience' (p. 148). The emotional
signal has an auxiliary role that increases the efficiency of the reasoning process
180 Fuller

FIGURE 10.8. An example of a potential 'emotionally competent' stimulus, linked to a

somatic markerof anxiety or fear.

but is not necessarily a substitute for it. However, when we immediatel y reject an
option that would lead to certain disaster, reasoning may be 'almost superfluous'.
Because emotional signals are body-related, Damasio labelled this set of ideas ' the
somatic marker hypothe sis'. Siovic et al. (2002) refer to a similar set of ideas as
' the affect heuristic ' . Throu gh learning, somatic markers become linked to stimuli
and patterns of stimuli. When a negative somatic marker is linked to an image of
a future outcome , it sounds an alarm (see an exampl e in Fig. 10.8).
What is compelling about the somatic marker hypothe sis and the affect heuris-
tic is the evidence cited in their support. Certain types of brain lesion specifically
exclude access to feelin gs associated with objects, events and scenarios. At the
same time they degrade decision performance: 'The powers of reason and the
experi ence of emotion decline together and their impairment stands out in a neu-
ropsychological profile within which basic attention , memory, intelligence and
language appear so intact that they could never be invoked to explain the patients'
failures in j udgement' (Damasio, 1994, pp. 53-54). Damasio (2003) has also out-
lined a plau sible and coherent neurological model which could sustain this entire
process. In addition, Siovic et al. (2002) and Loewenstein et al. (200 1) cite a num-
ber of experimental studies of decision making in normal individuals which clearl y
demon strate the interplay between emotion and reason, with the clear conclusion
that affect is essential to rational action.
In one such study cited by Siovic et aI., participants were asked to rate the
attractiveness of purchasing new equipment for use in the event of an airliner
crash-landing. It was hypothesised that saving 150 lives was a somewhat diffuse
 10. Motivational Determinants of Control in the Driving Task 181

positive outcome and would have a relatively weak positive effect. On the other
hand, saving 98% of something would be more convincingly good and would have
a much stronger positive effect. In one condition, participants were told that the
equipment would save 150 lives, which would otherwise be at risk. In a second
condition, they were told that the equipment would make it possible to save 98%
of the 150 lives, which would otherwise be at risk. It was found that support for
the purchase of the life-saving equipment was significantly higher in the 98%
of 150 condition than in the 150 condition. Support for the purchase was also
higher than in the 150 condition in a third situation in which participants were told
the equipment would make it possible to save 85% of the 150 lives that would
otherwise be at risk.
Now it might be suggested that explanation in terms of the somatic marker
hypothesis simply brings us right back to the threat-avoidance model, as pro-
posed in 1984 (Fuller, 1984), but with the concept of 'threat' being unpacked in
terms of its associated negative, punishing feelings. Damasio (1994) argues that
somatic markers are acquired through experience, under the control of an internal
preference system and under the influence of an external set of circumstances.
The internal preference system consists of 'mostly innate regulatory dispositions,
posed (poised?) to ensure survival of the organism' (p. 179). The external set of
circumstances includes

'events relative to which individuals must act; possible options for action; possible future
outcomes for those actions; and the punishment or reward that accompanies a certain option,
both immediately and in deferred time, as outcomes of the opted action unfold ... The
interaction between an internal preference system and sets of external circumstances extends
the repertory of stimuli that will become automatically marked' (p. 179).

This sounds remarkably like the description of an affect-conditioning history.

And indeed Damasio goes on to state: 'When the choice of option X, which leads
to bad outcome Y, is followed by punishment and thus painful body states, the
somatic marker system acquires the hidden, dispositional representation of this
experience-driven, noninherited, arbitrary connection' (p. 180). However the so-
matic marker hypothesis goes beyond this learning process. It asserts that affective
responses to presenting and anticipated stimuli not only inform response choice
(as in reinforcement theory) but also capture attention to pertinent stimuli and
prioritise their processing. In other words, affective responses have a direct effect
on cognitive operations.

10.2.5.1 Predictions and Speculations from the Somatic-Marker Hypothesis

One prediction from this hypothesis is that somatic marker strength should have an
impact on the distribution of attention over the traffic scene ahead of the driver. The
affective profile of the visual scene should be matched by the profile of attention
distributed over that scene. This selectivity might be operationalised in terms of
prioritisation and dwell time for particular stimuli in the environment, reflected,
for example, in patterns of visual fixations. In a field study of patterns of visual
182 Fuller

fixations by novice and experienced drivers, Underwood et aI. (2003) have shown
that on rural roads, two-fixation transitions by novices typically terminated in just
one zone, the road far ahead, whereas fixations by experienced drivers terminated
in five different parts of the scene. On a dual-carriageway, experienced drivers also
showed more extensive scanning, particularly in the horizontal plane. Underwood
et aI. interpreted this as evidence for greater sensitivity of experienced drivers to
prevailing traffic conditions. They concluded that the monitoring of other road
users is learned through experience and thus novices have relatively little ability to
switch the focus of their attention as potential hazards appear. The somatic marker
hypothesis offers the possibility that these learned differences in visual scanning
between novice and experienced drivers may be the result of learned affective
responses to events on the roadway, such as fast-moving vehicles merging from
both left and right in the dual-carriageway situation (Fuller, 2005b). Consistent with
this is the speculation by Loewenstein et aI. (2001), who suggest that age-related
differences in risk-taking may be 'affectively mediated', in particular, perhaps, by
differences in the vividness of mental simulations of consequences at the moment
of decision making.
Although linking of somatic markers to images is suggested to arise from a
process of learning, it is possible that affective responses to some events may be
unlearned reflex responses or that they are associated with a 'learning readiness' .
Put another way, it may be that there are universal somatically-marked stimuli.
Recognition of impending loss of control of a threatening situation might be one
such event that is, of course, highly pertinent to driver behaviour. Information
overload or an impossible response requirement may contribute substantially to
this recognition. Other unlearned but relevant affective responses, which might
at the same time also contribute to a feeling of loss of control, include the re-
sponses to a looming stimulus (rapid expansion of the retinal image of an object,
approaching collision; see, e.g., Schiff et aI., 1962; Bottomore, 1999; Franconeri
and Simons, 2003), to intense vestibular stimulation or g forces (e.g., Moro re-
flex; see, e.g., Goddard-Blythe, 1995) and to unexpected events (e.g., orienting
reflex). There may even be the equivalent of an inverse square law of affect inten-
sity in driving, with feeling intensity growing in proportion to the inverse square
root (or some other expression) of the time-to-line crossing or to collision, for
example.
A further prediction from the somatic marker hypothesis is that individual dif-
ferences in the affective response to particular scenarios should be associated with
different decisions in relation to those scenarios. Thus if we take speed choice
as an operationalisation of decision making, drivers who are more emotionally
reactive to road scenarios representing various degrees of threat (or impending
loss of control) should opt for lower speeds than drivers who are less reactive. If
there are stable individual differences in emotional reactivity (Larsen and Diener,
1987), this could mean that the same situation would ring alarm bells somewhat
differently for different individuals. Some may be relatively so 'deaf' that an im-
pending hazard has to be right on top of them before they are able to hear it, so
to speak. These persons would unwittingly be in a condition of delayed avoidance
 10. Motivational Determinants of Control in the Driving Task 183

(Fuller, 1984). Fujita et al. (1991) have shown that women experience negative
and positive affect more intensely than men. They asked 100 students to complete
the affect intensity measure (Larsen and Diener, 1987), which consists of 40 items
that measure how intensely participants feel emotions, yielding both a positive
and negative affect intensity score. Females scored higher on both positive and
negative affect intensity. Is it the case, then, that male-female differences in risk
taking are mediated by differences in emotional reactions to potentially hazardous
situations? Perhaps males crash more because they feel less.
Another individual difference factor that may be relevant in this discussion is the
tendency to seek enhanced external stimulation. Personality traits of extraversion
and sensation seeking are both considered to be constitutional characteristics of
the individual (Zuckerman, 1979) and both are associated with a preference for
enhanced levels of stimulation. Does this then mean that individuals with such
traits may be more likely to accept higher levels of somatic arousal and tolerate
more readily a driving situation where task demand is very close to capability, in
other words tolerate smaller safety margins? Research on individual differences
and accident involvement tends to support this prediction (Loo, 1979). Individu-
als high in sensation seeking are more likely to speed, overtake and adopt shorter
headways. They are also over-represented in traffic crashes (Jonah, 1997). Further-
more, Dahlen et al. (2005) have reported that degree of sensation seeking predicts
dangerous driving.
If the somatic marker hypothesis is correct in its implications for attention
capture by competent stimuli, the possibility presents itself that emotional re-
sponses arising from the unfolding road and traffic scenario may be drowned
out by or misattributed to other emotions or indeed, may even be suppressed or
extinguished (see concept of 'desensitisation' in Summala, 1986). For example,
feelings of anger may overwhelm the somatic marker indications, which would
otherwise inform decision making. Indeed is this the effect we are referring to
when we talk about rage being 'blind'? Deffenbacher et al. (2003) have shown
that high anger drivers are between 1.5 and 2.0 times more likely to engage in
risky driving, such as exceeding the speed limit and not wearing a seatbelt. Dahlen
et al. (2005), in a questionnaire study of 224 undergraduate drivers, found that the
propensity to become angry while driving predicted risky driving, minor losses of
vehicle control and loss of concentration. Levelt (2001) has reported that drivers
who say they are often irritated when driving also say that speeding is often the
consequence and Carbonell et al. (1997) found that anxiety combined with time
pressures can lead to engaging in dangerous manoeuvres (see review by Mesken,
2003).

10.3 Conclusions
Thus it is suggested here, albeit speculatively, that feelings of risk in driving may
arise from a sensitivity to changes which signal an impending loss of control and
that these feelings inform the driver's experience of task difficulty or workload.
184 Fuller

They (or in Damasio's sense, their underlying emotional substrates) may arise
naturally and spontaneously or through a process of learned association. They
have the power to direct attention to pertinent stimuli and to determine priorities
amongst response options. In short, they are integral to driver decision making,
even where they are so weak they are equivalent to what Slovic et al. (2002) call
'whispers of affect' . If we reinterpret the hypothesis of task-difficulty homeostasis
in terms of feelings of risk, the implication would be that drivers drive in such a way
as to keep feelings of risk below some threshold level (which may even be zero). If
the behaviour of other road users or the driver's own behaviour (such as an increase
in speed) should stimulate an increase in feelings of risk above this threshold level,
then the driver will take action to bring the level of felt risk back down, such as
by reducing speed. Only if the rewards of any supra-threshold feelings of risk are
compensated for by rewarding consequences of one sort or another will the driver
intentionally tolerate any increase above threshold. Nevertheless, increases in risk
may not be felt because of the swamping effect of other feelings or if felt they may
be attributed to events other than those related to the driving task. And experience
may not have provided sufficient learning opportunities to link particular potential
hazard scenarios to feelings of risk, that is, to provide the link to a somatic marker.
If we can accept all this as a working hypothesis and the evidence continues to
support it, then a whole new agenda for driver behaviour research emerges (see
Fuller, 2005b).
From the forgoing discussion it can be seen that the somatic marker hypothesis
has the potential to provide a unifying explanation for a diverse set of empirical
findings in the domain of driver behaviour, including our finding that feelings of
risk track ratings of driving task difficulty and speed almost perfectly. It also raises
a number of new questions regarding the role of affect and emotional condition-
ing in attention and decision making. This makes the experimental evaluation of
the somatic marker hypothesis of some importance in the contemporary research
agenda. It may be noted that one author has already begun to develop a model
of driver decision making based fundamentally on the somatic marker hypothesis
(Vaa, 2004, and this volume) and that Summala has proposed a thesis along simi-
lar lines (Summala, 2005, and this volume). The implication is that if we want to
understand driver decision making more clearly, we need to take into account not
just thinking but also feeling. Perhaps this is in part why decision making in safety-
sensitive industries, such as commercial aviation, has moved away from individual
decision making (reliance on somatic markers?) towards standard operating pro-
cedures (SOPs): Prescriptive rules for dealing with each contingency experienced.
In areas such as aircraft maintenance, where the affective consequences of inap-
propriate actions must be relatively weak, a reliance on SOPs must be especially
important for maintaining system safety. In this context any deviation from SOPs in
decision making is a matter of very serious concern (McDonald et al., 1999). Work
in this area also highlights the importance of avoiding conflict between SOPs and
what is perceived by the operative to be a better procedure - avoiding the tension
between formal and normal ways of working (Ward, 2005).
 10. Motivational Determinants of Control in the Driving Task 185

References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human
Decision Processes, 50, 179-211.
Armitage, CJ. and Conner, M. (2001). Efficacy of the theory of planned behavior: A meta-
analytic review. British Journal of Social Psychology, 40, 471-499.
Bern, S. and de long, H.L. (2006). Theoretical issues in psychology (2nd ed.). Sage Publi-
cations, London.
Bottomore, S. (1999). The panicking audience? Early cinema and the 'train effect'. Histor-
ical Journal of Film, Radio and Television, 19(2), 177-216.
Carbonell, EJ., Banuls, R., Chisvert, M., Monteagudo, MJ. and Pastor, G. (1997). A com-
parative study of anxiety responses in traffic situations as predictors of accident rates
in professional drivers. Proceedings of the Second Seminar in Human Factors in Road
Traffic, Universidade do Minho, Braga.
Dahlen, E.R., Martin, R.C., Ragan, K. and Kuhlman, M.M. (2005). Driving anger, sensa-
tion seeking, impulsiveness, and boredom proneness in the prediction of unsafe driving.
Accident Analysis and Prevention, 37, 341-348.
Damasio, A.R. (1994). Descartes' error: Emotion, reason and the human brain. Putnam,
New York.
Damasio, A.R. (2003). Looking for spinoza: Joy, sorrow and the feeling brain. Heinemann,
London.
Deery, H.A. (1999). Hazard and risk perception among young novice drivers. Journal of
Safety Research, 30(4), 225-236.
Deffenbacher, lL., Lynch, R.S., Filetti, L.B., Dahlen, E.R. and Oetting, E.R. (2003). Anger,
aggression, risky behavior, and crash-related outcomes in three groups of drivers. Be-
haviour Research and Therapy, 41,333-349.
de Waard, D. (2002). Mental workload. In R. Fuller and lA. Santos (Eds.), Humanfactors
for highway engineers. Pergamon, Oxford, pp. 161-176.
Elliott, M.A., Armitage, CJ. and Baughan, CJ. (2003). Drivers' compliance with speed
limits: An application of the theory of planned behavior. Journal ofApplied Psychology,
88(5), 964-972.
Evans, L. (1991) Traffic safety and the driver. Van Nostrand Reinhold, New York.
Franconeri, S.L. and Simons, DJ. (2003). Moving and looming stimuli capture attention.
Perception and Psychophysics, 65(7), 999-1010.
Fujita, F., Diener, E. and Sandvik, E. (1991). Gender differences in negative affect and well-
being: The case for emotional intensity. Journal of Personality and Social Psychology,
61,427-434.
Fuller, R. (1984). A conceptualisation of driving behaviour as threat avoidance, Ergonomics,
27, 1139-1155.
Fuller, R. (1991a). The modification of individual road user behaviour. In MJ. Koomstra
and J. Christensen (Eds.). Enforcement and rewarding: Strategies and effects. SWOV
Institute for Road Safety Research, Leidschendam, pp. 33-40.
Fuller, R. (1991b). Behaviour analysis and unsafe driving: Warning - Learning trap ahead!
Journal ofApplied Behaviour Analysis, 24, 73-75.
Fuller, R. (2000). The task-capability interface model of the driving process. Recherche
Transports Securite, 66,47-59.
Fuller, R. (2002). Learning to drive. In P. Barjonet (Ed.), Traffic psychology today. Kluwer
Academic Publishers, Dordrecht, pp. 105-118.
186 Fuller

Fuller, R. (2005a). Towards a general theory of driver behaviour. Accident Analysis and
Prevention, 37,461-472.
Fuller, R. (2005b). Driving by the seat of your pants: A new agenda for research.
In Behavioural research in road safety 2005. Department for Transport, London,
pp.85-93.
Fuller, R. and Farrell, E. (2001). Operation lifesaver assessment. RS 459. Project aLA,
NRA, Dublin.
Fuller, R. and Santos, 1.A. (2002). Psychology and the highway engineer. In R. Fuller and
J.A. Santos (Eds.), Human factors for highway engineers. Pergamon, Oxford, pp. 1-10.
Fuller, R., McHugh, C. and Pender, S. (2006). Task difficulty and risk in the determination
of driver behaviour. European Review ofApplied Psychology. In press.
Geller, E.S. (1998). Applications of behavior analysis to prevent injuries from vehicle
crashes. Behavior monographs: Cambridge Center for Behavioral Studies, Cambridge.
Goddard-Blythe, S. (1995). The role of primitive survival reflexes in the development of
the visual system. Journal ofBehavioural Optometry, 6, 31-35.
Harrison, W.A. (2005). A demonstration of avoidance learning in turning decisions at
intersections. Transportation Research Part F: Traffic Psychology and Behaviour, 8F,
4-5, pp. 341-354.
Hart, S.G. and Wickens, C.D. (1990). Workload assessment and prediction. In H.R.
Booher (Ed.). MANPRINT: An emerging technology. Advanced concepts for integrat-
ing people, machines and organizations. Van Nostrand Reinhold, New York, pp. 257-
300.
Hutton, K.A., Sibley, C.G., Harper, D.N. and Hunt, M. (2001). Modifying driver behaviour
with passenger feedback. Transportation Research Part F: Traffic Psychology and Be-
haviour, 4F, 4, 271-278.
Janssen, W.H. and Tenkink, E. (1988). Considerations on speed selection and risk home-
ostasis in driving. Accident Analysis and Prevention, 20, 137-142.
Jonah B.A. (1997). Sensation seeking and risky driving: A review and synthesis of the
literature. Accident Analysis and Prevention, 29, 651-665.
Larsen, R.J. and Diener, E. (1987). Affect intensity as an individual difference characteristic:
A review. Journal ofResearch in Personality, 21, 1-39.
Levelt, P.B.M. (2001). Emoties bij vrachtautochauffers. R-2001-14, SWay Institute for
Road Safety Research, Leidschendam, The Netherlands.
Loewenstein, G.F., Weber, E.U., Hsee, C.K. and Welch, N. (2001). Risk as feelings. Psy-
chological Bulletin, 127(2), 267-286.
Loo, R. (1979). Role of primary personality factors in the perception of traffic signs and
driver violations and accidents. Accident Analysis and Prevention, 11, 125-127.
Ludwig, T.D. and Geller, E.S. (2000). Intervening to improve the safety of delivery drivers:
A systematic behavioral approach. Journal of Organizational Behavior Management,
19(4),1-124.
MacCorquodale, K. and Meehl, P.E. (1948). On a distinction between hypothetical con-
structs and intervening variables. Psychological Review, 55, 95-107.
Makinen, T., Biecheler-Fretel, M.M., Cardoso, 1., Fuller, R., Goldenbeld, C., Hakkert, S.,
Sanchez Martin, M.C., Skladana, P., Vaa T. and Zaidel D. (1999). Legal measures and
enforcement. GADGET WP-5-Report, VTI, Espoo.
McDonald, N., Daly, C., Corrigan, S., Cromie, S. and Ward, M. (1999). Human-centred
management guide for aircraft maintenance. APRG, Trinity College, Dublin.
Mesken, 1. (2003). The role of emotions and moods in traffic. D-2003-8, SWOV Institute
for Road Safety Research, Leidschendam, The Netherlands.
 10. Motivational Determinants of Control in the Driving Task 187

Michon, J.A. (1989). Explanatory pitfalls and rule-based driver models. Accident Analysis
and Prevention, 21(4), 341-353.
Miller, G.A., Galanter, E. and Pribram, K.H. (1960). Plans and the structure of behavior.
Holt, Rinehart and Winston, New York.
Naatanen, R. and Summala, H. (1976). Road user behaviour and traffic accidents. North
Holland/Elsevier, Amsterdam and New York.
Oborne, D.1. (1995). Ergonomics at work (3rd ed.). Wiley, Chichester.
OECD - Road Transport Research (1990). Behavioural adaptation to changes in the road
transport system. OECD, Paris.
Payne, J.W., Bettman, J.R. and Johnson, EJ. (1993). The adaptive decision maker. Cam-
bridge University Press, Cambridge.
Ranney, T.A. (1994). Models of driving behavior: A review of their evolution. Accident
Analysis and Prevention, 26(6), pp. 733-750.
Reinhardt-Rutland, A.H. (2001). Seat-belts and behavioural adaptation: The loss of looming
as a negative reinforcer. Safety Science, 39, 3, 145-156.
Rumar, K., Berggrund, U., Jernberg, P. and Ytterbom, U. (1976). Driver reaction to a
technical measure: Studded tyres. Human Factors, 18,433--454.
Schiff, W., Caviness, J.A. and Gibson, J.1. (1962). Persistent fear responses in rhesus mon-
keys to the optical stimulus of "looming." Science, 136, 982-983.
Shugan, S.M. (1980). The cost of thinking. Journal of Consumer Research, 7, 99-111.
Slovic, P., Finucane, M.L., Peters, E. and MacGregor, D.G. (2002). Risk as analysis and risk
as feelings. Some thoughts about affect, reason, risk and rationality. Paper presented at the
Annual Meeting ofthe Society for Risk Analysis. New Orleans, Louisiana, 10 December,
2002.
Slovic, P., Fischoff, B., Lichtenstein, S., Corrigan, B. and Coombs, B. (1977). Preference
for insuring against probably small losses: Insurance implications. Journal of Risk and
Insurance, 44, 237-258.
Summala, H. (1986). Risk control is not risk adjustment: The zero-risk theory of driver
behavior and its implications. Reports 11: 1986. University of Helsinki Traffic Research
Unit, Helsinki.
Summala, H. (1997). Hierarchical model of behavioural adaptation and traffic accidents. In
T. Rothengatter and E. Carbonell Vaya (Eds.). Traffic and transport psychology: Theory
and application. Elsevier Science, Oxford, pp. 41-52.
Summala, H. (2005). Towards understanding driving behaviour and safety efforts. In L.
Macchi, C. Re and P.C. Cacciabue (Eds.), Proceedings ofthe International Workshop on
Modelling Driver Behaviour in Automotive Environments. European Commission, Joint
Research Centre, Ispra, Italy, 25-27 May, 2005. Office for Official Publication of the
European Communities, Luxembourg, pp. 205-214.
Taylor, D.H. (1964). Drivers' galvanic skin response and the risk of accident. Ergonomics,
7, 439--451.
Underwood, G., Chapman, P., Brocklehurst, N., Underwood, J. and Crundall, D. (2003).
Visual attention while driving: Sequences of eye fixations made by experienced and
novice drivers. Ergonomics, 46, pp. 629-646.
Vaa, T. (2003). Survival or deviance? A modelfor driver behaviour. TOI Report 666/2003,
Institute of Transport Economics, Oslo.
Ward, M. (2005). Contributions to human factors from three case studies in aircraft main-
tenance. Unpublished PhD thesis, University of Dublin, Trinity College.
Wilde, G.1.S. (2001). Target risk 2: A new psychology of safety and health: What works?
What doesn't? and why? PDE Publications, Toronto.
188 Fuller

Young, P.M., Clegg, B.A. and Smith, C.A.P. (2004). Dynamic models of augmented cog-
nition. International Journal ofHuman-Computer Interaction, 17(2),259-273.
Zajonc, R.B. (1980). Feeling and thinking: Preferences need no inferences. American Psy-
chologist, 35, 151-175.
Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal. Lawrence
Erlbaum Associates, Hillsdale, New Jersey.
11
Towards Understanding
Motivational and Emotional
Factors in Driver Behaviour:
Comfort Through Satisficing!
HEIKKI SUMMALA

11.1 Introduction
The early 'skill models' of driver behaviour and safety posited that the safety
of a driver is mainly determined by the level of his or her perceptual and motor
skills in relation to the task demands: a crash - a failure in driver performance -
occurs when task demands exceed driver capabilities (e.g., Blumenthal, 1968).
Consequently, improving driver skills and reducing task demands would make
traffic safer. Obviously, however, this early concept was too simple. A good piece
of counterevidence, among others, came from Williams and O'Neill (1974), who
showed that classified U.S. race drivers have more crashes per exposure than
average drivers - and also more speeding tickets. At least the advanced skills which
make those drivers competitive in race track did not save them from crashes, as
they obviously traded skills for speed on ordinary roads, too. Theorists and road
safety people indeed forgot that driving is a self-paced task and drivers themselves
do determine their task demands to a large extent (Naatanen and Summala, 1974).
The behavioural adaptation concept is now one of basic tenets in traffic psychology,
here defined by Summala (1996): 'the driver is inclined to react to changes in the
traffic system, whether they be in the vehicle, in the road environment, in road and
weather conditions, or in hislher own skills or states, and that this reaction occurs
in accordance with hislher motives' .
The safe completion of the trip is usually taken for granted by drivers. Therefore,
time goals, conservation of effort, maintenance of speed and progress, pleasure of
driving, all what Naatanen and Summala (1976) called extra motives, gain ground
in driver behaviour and cannot be forgotten anymore: they tend to push drivers
towards hazards primarily in terms of higher speed and shorter safety margins.
But what are the mechanisms which shelter us of crashes, of going too close to
hazards, and of exposing us to hazards?

IThis is an extension of the author's earlier paper presented in the symposium on traffic
psychology theories in the 3rd International Congress of Traffic and Transport Psychology
(see Summala, 2005).

189
190 Summala

11.2 Emotional Tension and 'Risk Monitor'

One line of thinking attempted to find one single motivational measure which
could explain driver behaviour, even including the moment-to-moment speed con-
trol. It started from the work of Taylor (1964). In two on-road studies he measured
galvanic skin responses in drivers who were driving in a wide range of roads and
road conditions. He found that GSR activity varied substantially by conditions but,
when controlled for speed, it was quite evenly distributed in time. He concluded
that it is the level of emotional tension or anxiety which guides the driver: 'Driving
is a self-paced task governed by the level of emotional tension or anxiety which
the driver wishes to tolerate'. Interestingly, Taylor probably interpreted his results
at least partly wrongly. Variation in speed and GSR activity in Taylor's data cor-
related with road type and road conditions such that lower speed and higher GSR
activity occurred on road sections with more side turnings and more other traffic
(Taylor, 1964). It can be assumed therefore that an important source of GSR was
motor activity rather than anxiety (see also Naatanen and Summala, 1976, p. 191).
However, Taylor's conclusion was quite influential on later theoretical work.
Wilde (1982), in his well-known and much debated model, proposed that drivers
tend to target a certain level of risk. The actual model is a simple homeostatic
(thermostatic) system with a traditional optimising decision model included in it,
but the model is best known from its safety prediction: Should any changes be made
in the traffic system, road users tend to maintain a certain target level of risk and,
therefore, safety level keeps approximately constant. It is obvious that this is not the
case, as already predicted by Smeed (1949) and confirmed by many big successes
in safety developments (see, e.g., Evans, 1991; Robertson and Pless, 2002). The
feedback loop from accident statistics (the knowledge of crash risk in a given
jurisdiction) in Wilde's model also seems quite inefficient to guide drivers' daily
task control and choices. Fuller (2000, 2004, 2005, see also Chapter lOin this book)
incorporates the concept of the target into his task-capability interface model.
He claims that drivers are sensitive to task difficulty and try to keep experienced
difficulty within a certain margin in a homeostatic loop. Vaa (2004, see also Chapter
12 in this book) lists several candidates for drivers' targets. Drivers may target for
a certain arousal level, in all of its varieties, and sensation, pleasure, security,
workload, avoiding violations or even non-compliance of the rules. Finally, while
applying Damasio's model of emotions and feelings, he proposes that the pacing
factor is target feeling or best feeling, and the body is the risk monitor (Damasio,
1994).
But to what degree dynamic driver behaviour can be explained with one 'mo-
tivational', targeted control measure which drivers tend to adjust? And, could it
be possible to operationalise such a measure to test it adequately, and to apply in
automotive and transportation research and development?
Naatanen and Summala (1974, 1976) adopted quite a different position. In line
with Taylor (1964), they claimed that driving is a self-paced task. They proposed
that the task difficulty level is determined by the drivers according to their motives.
Drivers are not only able to compensate for changes in the degree of difficulty of
 11. Towards Understanding Motivational and Emotional Factors 191

(a) (b)

FIGURE 11.1. (a) Humans can be seen as having different zones (intimate, personal, social)
around them (Hall, 1966), with thresholds which trigger approach or avoidance response,
depending on the approaching person (or animal or object). (b) When in motion, more space
is needed in front to avoid collisions with objects. The safety zone (Gibson and Crooks,
1938), whether in motion or not, can also be seen as a comfort zone, with no threat, risk, or
discomfort felt (cf. 'zero-risk model', Naatanen and Summala, 1976; Summala, 1988).

traffic situations by modifying their efforts (attention, vigilance), but they also
determine the nature and degree of the difficulty of these various situations and
their current task (Naatanen and Summala, 1976, p. 35). This is the prerequisite
for the fact that many safety measures have not led to expected results.
In sharp contrast to Taylor's and Wilde's models, they proposed a threshold
model with three major starting points:

(1) 'Subjective risk monitor' for present or anticipated fear is the major inhibitory,
limiting mechanism in driver behaviour.
(2) Drivers' goals and excitatory motives push them towards the limits, e.g., to-
wards higher speed if not otherwise restricted.
(3) Safety margins are in a key position in driver's task control (see the next
section).

In their control flow model, Naatanen and Summala (1974) postulated a moti-
vation module which produces desired actions or tends to keep pace (to maintain
progress). Action is continuously monitored by the subjective risk monitor which,
given a certain threshold is exceeded or anticipated, alerts and takes a role in deci-
sions. Drivers' goals and so-called extra motives, either arisen in traffic or brought
from outside it, got an important role in Naatanen and Summala model.

11.3 Safety Margins and Safety Zone

Another line of thinking comes from two sources, from the so-called proxemics
approach (Hall, 1966) and from an early work of Gibson and Crooks (1938) on
safety zone (see Fig 11.1).
Hall (1966) described people having different zones around them, depending on
who is approaching. The intimate zone, with full contact, is reserved only for closest
people - family members and best friends - while personal, social and public zones
are applied for less familiar people and social situations. These zones translate to
critical distances, thresholds which trigger approach or avoidance response, or
192 Summala

flight or fight response. In public space an approaching person's appearance and

his or her perceived or interpreted intentions influence on feelings of comfort and
safety, that is, the threat the approaching person represents. In that sense, humans
have safety or comfort zone around them in all environments, with strong emotional
characteristics: intrusion into this zone arouses discomfort.
When a human being is in motion, he or she must reserve additional margins,
especially ahead of him or her, to avoid colliding with obstacles: The faster the self-
motion the more space is needed in front. As noted by Rumar (1988), a Swedish
engineer Sylwan (1919) already described the physical and psychological space
which different traffic units require on streets, from a slowly moving pedestrian to
a fast car. Two decades later, Gibson and Crooks (1938), in their excellent analysis
and series of drawings, demonstrate how roadway, obstacles and other road users
modify this space - safety zone. They also implied that safety zone - and stopping
distance within it - is an objectively measurable concept:

Phenomenally, it is a sort of tongue protruding forward along the road. Its boundaries are
chiefly determined by objects or features of the terrain with a negative 'valence' in percep-
tion. It is not, however, merely a subjective experience of the driver. It exists objectively as
the actual field within which the car can safely operate, whether or not the driver is aware of
it. It shifts and changes continually, bending and twisting with the road, and also elongating
and contracting, widening and narrowing, according as obstacles encroach on it and limits
its boundaries. (Gibson and Crooks, 1938, p. 121)

The safety zone, or safety margins which road users must keep around them,
can be expressed, measured, and are functional in both space and time. We have
to drive a car through gaps, in space or time, and to keep distance to other ve-
hicles and to pedestrians, bicyclists and road-side obstacles. This distance from
crash is essential metrics in everyday control of safety. On a two-lane road, for
example, an almost certain death lurks at a distance of 2 to 3 m when one meets
a heavy vehicle (even a car) and every driver must take care of keeping a suf-
ficient margin. If the distance - safety margin - is not wide enough, we feel
uncomfortable.
However, car control is also extensively based on time margins such as time-
to-collision and time-to-line-crossing. Therefore, keeping the car in a lane is not
a simple tracking task nulling the error from the mid of the lane or the intended
trajectory (cf. Donges, 1978). The latter model might be relevant in 'active driving' ,
with all attention focussed on driving and optimal lane control. In normal everyday
driving, instead, a lane should rather be conceived as a tube with a lot of tolerance
and time to correct path within it. As proposed by Godthelp et al. (1984), time-
to-line-crossing is a relevant measure, referring to the time until the car drifts out
of the lane if not corrected through steering. It is the measure which sets the time
limit - the threshold - for drivers' path correcting.
Time-to-collision or time-to-contact is a central control measure when we con-
tinuously adjust distance to a braking vehicle in front of us (Lee, 1976), or start
braking when approaching an obstacle or a stop line at a crossing (van der Horst,
1990). It defines time distance to a crash, or to the moment that braking must be
 11. Towards Understanding Motivational and Emotional Factors 193

started to avoid the crash. As proposed by Lee (1976), the optical variable r alone
gives a direct approximation of the time to collision, and its first-time derivative
(tau dot) provides a feasible strategy to control braking when kept within a certain
range. There is indeed a lot of evidence that the human perceptual system is well
equipped to accurately estimate time to collision, at short distances at least when
optic resolution of the looming object is sufficient. This means that at any moment
we know when we will crash with a car or an obstacle we are approaching to.
However, this information does not yet reveal whether we are able to stop before
the obstacle, or when we should start braking to avoid the crash.
Quite recently, Fajen (2005) showed that performance in stopping in front of an
obstacle also depends on global optic flow and edge rate (providing speed infor-
mation)' and that information is needed (along with tau) to tell whether stopping
is possible with available braking force. In other words, this information - open
to continuous calibration of maximum deceleration level - provides us with an
estimate of the action boundary, that is, the limit above which stopping is no more
possible. Fajen (2005) also returns to Gibson and Crooks (1938): it was already
in their paper that, in defining the safety zone, they made a distinction between
trajectories which are available and the ones which not. It was a beginning of
the later development of Gibson's concept of affordance - what the environment
affords to the human.
It should be added that (mainly vision-based) space and time margins are not
of course the only ones to determine and to set limits for driver behaviour. For
example, while time-to-line-crossing has a role when approaching a bend, lateral
acceleration is a marked factor and a marked source of proprioceptive and kines-
thetic information in curves and steering manoeuvres (Reymond et al., 2001).
Available friction in a curve can also be seen as a safety margin measure (Rumar
et al., 1976; Summala and Merisalo, 1980; Wong and Nicholson, 1992), indicating
a distance to a loss of grip and control. At low friction and low speed in a steep
off-ramp drivers typically drive close to the threshold of loss of control, in terms
of available friction, while safety margins grow with friction (when slipperiness
decreases) and speed (Summala and Merisalo, 1980). It suggests that available
friction margin is far from a simple distance measure. Drivers may misperceive
the slipperiness at approach phase - in wintertime lower volume ramps are often
more slippery than motorway lanes - but they may also accept a smaller margin to
grip loss at slow speed when the chances to regain control are bigger (cf. Brown,
1980) and consequences of loss of control and running off the road smaller. Some
drivers - young men especially - may also intentionally make their car skidding
in more or less good control.

11.4 Available Time, Workload and Multilevel Task Control

It is essential to note that time safety margins have an important feature: they imply
a concept of available time. Available time determines brake reaction latencies
as well as time sharing while driving, among other things. The timing of brake
194 Summala

response in front of an unavoidable obstacle therefore depends on the time-to-

collision (or on time to the moment that braking is to be started to provide a
smooth and comfortable stopping). Therefore, the brake reaction time of unalerted
drivers in real-life conditions depends on the urgency of the situation, however
such that there is a substantial variation due to drivers' attentiveness, age and
individual response style among others (Summala and Koivisto, 1990; Summala,
2000).
For an experienced driver, especially in routine open road driving, time is an
abundant resource which can be used for many kinds of in-car activities. As ex-
pected, in terms of time-to-line-crossing, a wider road means more time and, ac-
cordingly, drivers allow longer glances and more time for subsidiary tasks (Wikman
et al., 1998; Wikman and Summala, 2005). Similarly, less time is available - and
used - for an additional task at curves and at higher speeds, when more time is
needed to update position information, to predict the course, and even to plan the
next step in a subsidiary task (Wikman et al., 1997). On a curvy road at high speed
drivers have to attend to the road and steering entirely, and subsidiary tasks typ-
ically drop out or slow down when available time (and spare capacity; see, e.g.,
Harms, 1991) diminishes.
The concept of available time is closely related to workload felt by drivers. It
is the other side of task complexity, the information to be processed in a time
unit. The more complex the task (e.g., traffic environment), given same speed,
the higher the workload. By adjusting speed, however, the drivers can reduce the
information processing rate and provide themselves with more time, make their
task more controllable and less loading, simply speaking, less difficult.
Among many dimensions of workload (e.g., Hart and Staveland, 1988), time
and time pressure are critical always when a human is in motion, and especially at
speeds of vehicular traffic. This is in line with Hancock and Caird's model, which
predicted that mental load grows as effective time for action decreases (Hancock
and Caird, 1993). On the other side, sufficient time and adequate time margins
imply the feeling of control, comfort and safety. This is also the essence of the
expertise which practice brings for drivers during the first tens of thousands of
kilometres, when car control skills get more automated and require less attention,
and improved anticipatory traffic control skills allow more time to prepare for
potential hazards.
Accordingly, the hierarchical model of behavioural adaptation (Summala, 1997)
predicts that available time and related workload mediate between operational and
tactical, even strategic levels of driving. Figure 11.2, outlining the role of time in
traffic behaviour, proposes that mobility and other goals of driving influence on
trip decisions and target speed planned to complete the trip. This target (desired)
speed further affects on safety margins and modifies and generates manoeuvres,
for example, overtaking a slower vehicle on highway or weaving in and out for a
faster lane and progress in city.
A control loop from operational to upper levels is assumed on the basis of safety
margins and available time. Given a certain speed and certain complexity of the
environment, if a driver cannot complete all operational subtasks (in the control
 11. Towards Understanding Motivational and Emotional Factors 195

AVAILABLE TIME and (expected) SPEED OF TRAVEL

L MOBILITY NEEDS
and possibilities

strategic level
~
TRIP DECISIONS

~ SHORT
OF
TIME:
TARGET SPEED increased
tactical level

MANOEUVRING
/" mental
load,
modifies
overtaking higher-
~~~~sing management ~ level
goals

SAFETY
operational level
MARGINS
time margins
PLENTY OF
TIME:
highly
automated,
effortless
easy
monotonous
task

TIME SHARING REDUCED

in-car tasks AROUSAL
divided attention sustained attention
problems problems
OVERLOAD UNDERLOAD

FIGURE 11.2. The role of time is essential when a human is in motion and making travel
decisions. Time margins and available time for action, through mental load, also mediate
between different levels of driving (adapted from Summala, 1997).

of car and traffic) at this speed, within available time margins, he or she feels high
mental load and
either slows down to get more time, to reach the feeling of control and an acceptable
(or comfortable) level of workload; or
continues in the unstable zone, feels increased mental load, and may experience
sudden overload problems.
However, if a driver repeatedly or continuously experiences certain conditions
too loading, pressing, or difficult, he or she starts to avoid them. For example,
196 Summala

elderly drivers with slower performance (and lower contrast sensitivity and in-
creased glare discomfort) start to avoid driving in the dark, in heavy traffic and in
winter time, etc.
On the other side, if a driver has plenty of time, like on a high-standard road
with a fairly low speed limit, he/she is inclined to perform secondary tasks; or
is at risk of getting bored and tired, both of which mean unstable performance
of the driver/vehicle system. Secondary tasks, while driving, require reliable time
and attention sharing and easily lead to delayed detection of relevant information,
overload, and increased risk. Underload and lowered arousal level also result in
delayed responses and attention lapses, if not falling asleep.
Available time incorporates many essential features like task (environment)
complexity, mental load and stress. It is, by definition, always a factor when humans
are in motion. It is also a factor when road users are doing decisions while stopped
in traffic, in gap acceptance for example, when drivers are crossing or entering a
priority road. In that case the traffic flow on a main road actually means making
a selection from a number of closing time margins, and the situation may be very
loading.
The key issue in the model is that time margins (in the meaning of distance from
crash or from the normal or 'last' response threshold) are very basic measures for
humans, with a strong affective component if certain limits are violated. Available
time in tum defines what we can do in each situation, how loading we feel a
situation, and therefore it provides both cognitive and affective mechanism for
control. Hollnagel (2002) and Hollnagel and Woods (2005) similarly gave the
concept of available time a marked position in their Extended Control Model,
such that time and control are closely intertwined and mediate between different
hierarchical control levels (see also Chapter 4 in this volume).

11.5 Safety Margins, Affordances and Skills

We see that time (and space) margins have a double role in driver behaviour
determination. On one hand, they set (and show) the limits to how close to a hazard
we can go and, on the other hand, they provide (and show) us with opportunities
to promote our goals and motives.
Time and space are actually resources which the traffic system affords, and we
more or less actively look for time and space (or, to be more accurate, spatiotem-
poral) slots to promote mobility and progress. We look for sufficient - affordable,
safe, comfortable - gaps in main road flow to cross the road, in the oncoming
flow to afford overtaking a slower vehicle, in an adjacent lane to change lane, and
empty (or faster) lanes to promote progress in city. Time and space slots can be
seen as opportunities which dynamic traffic environment affords - affordances in
Gibson's terminology (Gibson, 1979).
It is important to note that what the environment affords is necessarily re-
lated to one's abilities, skills and physical characteristics (e.g., Greeno, 1994).
Warren (1984) nicely showed that 'climbability' of stairs of different height is
 11. Towards Understanding Motivational and Emotional Factors 197

directly dependent on the length of the knee, and Warren and Whang (1987)
showed that an affordable aperture for walking through without body rotation de-
pends on shoulder width. Available space in relation to the shoulder width there-
fore determines the action which is most comfortable and efficient, as well as
the transition point between two choices, unrotated and rotated pass. The affor-
dances of the car/driver system are similarly related to the 'body size' of the
vehicle, but also on the driver's ability to estimate car dimensions relative to the
available gap and to steer through it. Similarly, a gap between vehicles in the
oncoming flow is sufficient or not for overtaking a slower vehicle in front, de-
pending on its size, vehicles' speeds, the own car's acceleration and the driver's
estimation skills. Therefore, the skills acquired during the driving career essen-
tially affect on what is affordable. This is also the essence of the process where
novice drivers extend the realm of the possible operations they will attempt while
learning to know dynamic, spatiotemporal limits of the driver-car unit amid of
traffic:

Every decision made, every action taken, every traffic situation one is exposed to, provides
some kind of feedback to the driver. It is presumably this feedback that is the really efficient
driving teacher. As to the aspect of the safety of different actions, decisions and situations, the
related feedback is presumably received in the form of subjective time and space separating
the driver from an accident; hence the driver can test different kinds of operations, driving
manners, etc. against these subjective dimensions and thus finally develop rather permanent
criteria for different kinds of actions, decisions and traffic situations. It is proposed that
these criteria are subject to a continuous back-and-forth change, at least to a minor extent,
through the whole driving career. (Naatanen and Summala, 1976, pp. 87-88)

Among recent conceptualisations of the driver's choices between options,

Groeger (2000), in his differential analysis, gives a detailed model of how drivers
manage with implied goal interruptions, appraising them, and planning and im-
plementing actions to handle them, along with a multitude of factors that explain
interindividual variance in these performances. Goodrich et al. (2000) first defined
a skill as a learned sequence of human activities, whether it be simple or com-
plex, and then proposed that humans map environmental cues for a given task,
and implement a skill when appropriate. Goodrich et al. fix affordances to skills
such that the (cognitive) skills provide affordances for rational behaviour, with
different attractive and repulsive potentials (p. 94): 'Skills whose affordances are
compatible with top-down goals induce an attractive potential commensurate with
their likely usefulness .... However, in addition to task specific goals there are
also context-dependent constraints on the efficiency of these skills, and these con-
straints induce a repulsive potential commensurate with their likely inefficiency.'
The choice of an option then depends on the trade-off between its attractive and re-
pulsive potentials (benefits and costs): this comparison defines whether the option is
satisficing.
We see here again a certain resemblance with Gibson and Crooks (1938), who
joined negative valences to those trajectories which are not available for a driver,
and positive valences to the ones which are available. In Damasio's (1994) terms,
198 Summala

we could say that in the potentially hostile road environment the options on the
road obviously are tagged with either negative (risky) or positive (goal-directed)
somatic markers. These markers then precede rational evaluation and facilitate and
speed up the choice. In dynamic time-limited situations like driving, fast affective
heuristics must have a big role (Finucane et aI., 2000). We could even say that safety
margins have such a role (see already in Summala, 1988), telling what choices are
affordable and what are not, and when the situation is going out of the control and
needs an appropriate reaction. In line with Naatanen and Summala's model, given
certain environmental information through perception and expectancies it triggers,
motives feed desired actions while 'risk monitor' blocks implementation if fired
through perceived or anticipated threat.

11.6 Towards Unifying Emotional Concepts

in Routine Driving
Emotional tension, task difficulty, or best feeling has been proposed (see above)
as critical motivational or affective control measures which guide driving. Rather
than a single (affective) control measure, there are many processes involved, and
a unifying concept is sought here which should cover all major determinants of
driving. It is also an attempt to outline what might be the role of weak emotions
in driving. Although traffic psychologists have mostly considered strong emo-
tions such as anger or fear, however, weaker, less aroused emotions may have
a much bigger role in everyday driving, and highly probably even in everyday
crashes.
Safety margins (and safety zone) are necessarily the primary and very basic
control measures in on-road driving and, as proposed above, both continuous task
control and choices at tactical and strategic levels can be defined using them.
They both set the limits to how close to a hazard we can go and provide us with
opportunities to promote our goals and motives. As proposed above, time margins
also provide a mechanism for explaining mental load while in move.
In explaining driver behaviour we also have to conceptualise motives and goals
which push drivers towards hazards - towards shorter safety margins. This is not
enough yet however. We have to incorporate rules, social norms, vehicle and road
system (in so far as not already included in the control of safety margins, in what
is affordable on a certain road with a certain car). I recently proposed what I called
'multiple comfort zone model' to incorporate all relevant factors into one general
framework (Summala, 2005).
This is not one-dimensional target model. Instead, a few functional control
variables are defined (not exhaustively) which drivers are assumed to keep within
an acceptable range, in the 'comfort zone', such that this process determines not
only, say, the speed level in normal routine driving or a decision whether to overtake
a slower car, but also results in a general 'comfortable' or 'best feeling' state. It
may be seen as a target state, given a certain trip and conditions, but it is rather an
 11. Towards Understanding Motivational and Emotional Factors 199

output of many cognitively definable processes all of which can be included under
the shared umbrella concept of comfort.
The factors to be kept within the 'comfort zone', to exemplify normal driving
situations, are proposed below, both including inhibitory and excitatory ones. Note
that for each main factor, there are specific models to explain cognitive processes
involved.

11.6.1 Safety Margins - To Control and Survive

The 'comfort zone' implies sufficient space and time margins around the driver,
that is, to road edges, obstacles, other vehicles and, finally, to a crash. Safety
margins are understood as being the major tool for survival and the major control
variable, and they also provide efficient feedback in the learning phase. Sufficient
margins are needed by drivers to feel safe and comfortable while driving, with no
excessive mental load, also implying that they feel able to manage with all subtasks
within available time.
A threshold model for safety margin management is assumed. It is assumed
that drivers normally feel full control over the task and no risk while driving. At
certain (inherent and learned) thresholds of safety margins, corrective steering and
speed adjustment is triggered. Too short time or space distances make drivers to
feel uncomfortable, and they do not tolerate it at least without a very strong motive
to continue.
It is to be noted that the ordinary language use of 'comfort' also corresponds
very well to the essence of safety margins: a person may be too close to a danger
(or cliff edge, road edge, a truck, etc.) for comfort. Several researchers have indeed
used the comfort criterion or scale in their experimental research while assessing
a limit for normal, comfortable, acceptable driving. A close connection of safety
margins to the concept of comfort was already shown by Godthelp (1988), who
instructed drivers in open road driving to correct their path only at the moment
when it could still be corrected comfortably to prevent a crossing of the lane
boundary. This resulted in a constant time distance (time-to-line crossing) at the
moment of decision over a broad range of speeds. Ohta (1993) asked drivers to
follow a car under four instructions in his on-road study, including, 'follow at
a distance which you feel most comfortable; 'approach until you begin to feel
the distance dangerous'; 'follow at the minimum safe distance'; and 'follow at a
distance which you feel to be neither too far not too close'. On the basis of his
data and Hall's (1966) proxemics concept, he defined a danger zone (headway up
to 0.6 s), critical zone (up to 1.1 s), and comfortable zone (1.1 to 1.7 s, mean 1.4
s). Interestingly (cf. below), he concluded that longer headways, rather considered
as pursuit distances, are uncomfortable for steady state following because they
are against the social norm. For other relevant studies, see also, for example, van
Winsum and Heino (1995), De Vos et al. (1997), Taieb-Maimon and Shinar (2001).
Other things being equal, motives typically push drivers towards the safety
margins threshold but, typically, they keep at a comfortable margin from it. A
violation of the safety/comfort zone alarms, and a sudden unexpected violation
200 Summala

frightens: this is the affective 'subjective risk' monitor proposed by Naatanen and
Summala (1974, 1976).

11.6.2 Vehicle/Road System - To Provide Smooth

and Comfortable Travel
The modern cars are silent and go smoothly at speeds normally used. Poorly
balanced tires make the car shaking at high speeds, however, and gravel, snow or
ice on road makes control much less confident and comfortable. Thermal comfort,
seat comfort and vibration are a widely studied area with advanced models on what
is comfortable (e.g., Jiang et al., 2001; Gameiro da Silva, 2002). Glare discomfort
has long been used as a criterion in vehicle headlight studies (Olson and Sivak,
1984; Sivak et al., 1991; Theeuwes et al., 2002), while visual comfort is a general
concept (and scale) for texts, pictures and visual display units (see, e.g., Roufs and
Boschman, 1997).
Drivers also appear to have certain thresholds for stopping deceleration, indica-
tive of discomfort. Thus, while approaching a signalised intersection, drivers tend
to pass through if the required deceleration exceeds 3 to 3.5 ms- 2 at the time of
yellow onset (Baguley, 1988; Niittymaki and Pursula, 1994; van der Horst and
Wilmink, 1986). The motion change (jerks or the acceleration's time derivative)
has been used as a passenger's comfort metrics in public transportation, with the
threshold of 2 ms ? (Canudas-de-Wit et al., 2005).

11.6.3 Rule Following - To Avoid Sanctions

Rule following is a major pacing factor when the speed limits restrict speed rather
than the infrastructure. The 'comfort zone' implies no concern of getting fined.
This is a simple mechanism which incorporates speed limits into speed adjustment
by drivers. The cognitive analysis may indeed include optimisation with estimates
of police enforcement, for example. However, strictly keeping to speed limits or
to the announced sanction limit or 'driving with others' are those simple strategies
which guarantee comfortable mood. Compliance with rules means not only the
law but also social norms, which may differ a lot from what is defined in the law.
Therefore, actual speed level on the road affects drivers' target speeds beyond
speed limit (e.g., Haglund and Aberg, 2000), while drivers' perception of other
people's speed may be biased and tend to amplify the social norm effect (Connolly
and Aberg, 1993).
Some drivers appear to have a rather consistent intention to speed, also related to
higher crash risk, which however seems to indicate general social deviance rather
than, for example, underestimation of consequences (Lawton et al., 1997). A bulk
ofresearch has focused on analysing interindividual factors behind rule breaking in
the context of Theory of Planned Action (Ajzen, 1991). However, in daily choices
and ordinary driving, habit seems to be a stronger determinant of behaviour than
intention (Ouellette and Wood, 1998; Verplanken et aI., 1998).
 11. Towards Understanding Motivational and Emotional Factors 201

11.6.4 Good (or Expected) Progress of Trip -Mobility

and Pace/Progress
Good progress of the trip is a major mobility factor. Keeping in the 'comfort zone'
implies that travel progresses as expected, while tendency to maintain progress -
speed and fluent progress - may grow a strong motivational force such that de-
celeration is felt punishing and keeping the speed and pace represent the 'comfort
zone'. Sensory and cognitive adaptation even tends to move the current speed up-
wards little by little. Therefore, these mobility-related goals should be conceived
as excitatory ones. They push drivers towards rule breaking and critical safety
margin thresholds.

11.7 Comfort Through Satisficing

A few major factors were listed above, relevant for a driver when he or she controls
a car in traffic in normal everyday driving. It is hypothesised that drivers normally
keep each of them within a certain range (or above certain threshold), in a 'comfort
zone' . This mechanism results in a comfortable state, or best feeling in the sense of
Damasio (1994) and Vaa (2004). The process can be called satisficing, in contrast to
optimising, according to the well-known principle of Simon (1955), who proposed
that the option is chosen which first fulfils a certain aspiration level on a few criteria.
Emotions can be practically located on two dimensions, pleasure and intensity.
Comfort is here thought as a general mood, or emotion which is pleasant but
not especially aroused, tense, or activated. Fig. 11.3 illustrates 'driving moods'
in a schematic map of core affects (Russell and Barrett, 1999). Routine, daily,
comfortable driving ranges from contended to relaxed or calm, while reduced
arousal level (task induced, circadian or sleep deprived) makes driving drowsy
and unpleasant.
Comfort is pleasant, by definition, and indeed riding comfort is sometimes
equalled to pleasure (e.g., Canudas-de-Wit et al., 2005). A driver can experience
pleasurable feelings in normal, safe, rule-following driving, for example, due to
feeling of control, and when travel goes on fluently, without jams, delays or extra
decelerations due to slow drivers ahead. Ride comfort of the car undoubtedly affects
to such feelings. At best, very normal safe driving provides a flow experience of
optimal control.
Hedonistic motives and pleasure indeed have long acknowledged as impor-
tant determinants of driving (e.g., Black, 1966; Naatanen and Summala, 1976;
Rothengatter, 1988) which, however, easily slip to hazardous behaviour. Speed
tends to grow, and even normal comfortable driving may have negative effects on
safety. Keeping the pace tends to elevate the threshold of slowing down for safety,
and higher speed even forces to selective visual search which ignores less probable
risks (e.g., Rasanen and Summala, 2000; Summala and Rasanen, 2000). Driving
for pleasure may mean sensation seeking and 'high' from speed, acceleration, and
close margins, and competitive tendencies. Other extra motives, additional to pure
202 Summala

Urgenthurry Reactive
Beginner
ACTIVATION
tense Competitive
alert
Frustrated, nervous excited
aggressive
stressed

upset

sad

depressed
relaxed
lethargic Comfortable,
fatigued
calm routine

DEACTIVATION

Drowsy

FIGURE 11.3. Driving moods as projected on the two-dimensional schematic map of core
affects (Russell and Feldman Barrett, 1999). Routine, daily, comfortable driving ranges
from contended to relaxed or calm.

mobility goals, such as hurry, social pressure (in car and by other drivers), self-
enhancement, competition, thrill seeking, may also take control. Stronger emo-
tions arise when driving becomes competitive, or reactive in heavy traffic, while
hurry and frustrated aggressive driving exemplifies aroused (activated) emotions
on the side of displeasure (Fig. 11.3). Even a thought that the roadway might
be slippery due to black ice makes driving tense and unpleasant, at least for a
while.
However, it is claimed here that normal everyday driving is largely habitual
activity, aiming at keeping within comfortable limits, with no feelings of risk, or
anxiety, or discomfort. The output is, then, a fairly stable emotional state (which,
unfortunately, every now and then is disturbed by hurry, frustration, aggression
and other strong effects). Our cognitive system provides fairly good information
on where the limits of 'safety zone' are (even if it often tends to be biased) while
the emotional system essentially contributes and warns about 'wrong' choices or
when limits are being approached. This is well in line with weak emotional signals
proposed by Damasio (1994; for more detail, see also Chapters 10 and 12 in this
volume).
It is interesting to see how strongly emotions have now intruded into modelling
human (risky) choices. Damasio's (1994) book and subsequent experimental work
(e.g., Bechara et aI., 1997) has attracted much enthusiasm. Also the long tradition of
risk perception research, in the meaning of evaluating risks of different activities
 11. Towards Understanding Motivational and Emotional Factors 203

(e.g., Fischhoff et al., 1978) - has now more explicitly welcome the emotional
component in risky decisions, 'risk as feeling' (e.g., Slovic et al., 2004). However,
early 'risk theories' in driver behaviour already assumed a marked emotional
component in driver task control. Taylor (1964) indeed launched the 'emotional
tension' as a critical measure, and Naatanen and Summala (1976), while describing
the function of 'risk monitor' , considered risk as feeling:

Hence, the general view of the road users' behavioral dynamics was advanced that his
behavior is continuously pushed by his (other-than-safety) motives in their direction. For
example, a driver in a hurry wants more and more speed, but that somewhere along this shift
toward those behavioral forms which give more and more satisfaction to his (excitatory)
motives (which, simultaneously, are also in general more dangerous), the' Subjective Risk
Monitor' becomes activated and this development is usually soon stopped. It was suggested
that under 'normal', 'relaxed' driving motivation, only those decisions are made which are
not associated with subjective risk (at the moment of decision) and that when subjective
risk is felt in some actual driving situation, the behavior is changed so as to eliminate the
source of this feeling. (Naatanen and Summala, 1976, p. 221)

11.8 'Go to the Road': Need of On-Road Research

However, it is not time to carry away the cognitive basis of driver task and risk
control, but rather to remind that the cognitive and emotional systems cooperate
closely intertwined. Much confusion has originated from the misunderstanding of
their roles in interpreting different kinds of evidence in driver behavior analysis.
Risk estimates 'from arm chair' (see McKenna, 1982; Summala, 1985), while
filling questionnaires, pushing buttons in laboratory experiments, or even while
driving in simulators, do not grasp real emotional contents of on-road driving.
Even Fuller (2005, Fig. 5; and this volume) appears to misinterpret the results from
his recent laboratory experiment in which participants looked at videoclips taken
from a highway at different speeds. 'Estimated task difficulty' and (conscious,
rational) 'risk experience' indeed correlated perfectly in his data - even to the
degree that they seemed to be inseparable and semantically confused measures
drawn from rational evaluation. However, when the presentation speed increased,
the threshold where the estimated crash risk deviated from zero obviously equalled
to the threshold for not being able to manage at higher speed: '95% of the sample
would be uncomfortable driving at a speed at which there was some estimated risk
of crashing' (Fuller, 2005, p. 469). Even if only based on a cognitive evaluation
of the task in a laboratory and obviously not capable of triggering any emotional
response (with no true hazard, no game or no social pressure included), this result
appears to support the threshold model. We have indeed to assume that mobility-
related and other excitatory motives push drivers towards a threshold such that the
task difficulty grows and task demands follow capabilities at a certain margin.
Finally, when the role of motives and emotions in driver behavior is now gener-
ally acknowledged, theorising needs testable hypotheses and their tests in real-life
driving.
204 Summala

References
Ajzen, L (1991). The theory of planned action. Organizational Behavior and Human Deision
Processes, 50, 179-211.
Baguley, C.1. (1988). 'Running the red' at signals on high-speed roads. Traffic Engineering
and Control, 29, 415-420.
Bechara, A., Damasio, H., Tranel, D. and Damasio, A.R. (1997). Deciding advantageously
before knowing the advantageous strategy. Science, 275(5304), 1293-1295.
Black, S. (1966). Man and motor cars. Seeker & Warburg, London.
Blumenthal, M. (1968). Dimensions of the traffic safety problem. Traffic Safety Research
Review, 12,7-12.
Brown, LD. (1980). Error-correction probability as a determinant of drivers' subjective risk.
In D.1. Oborne and lA. Levis (Eds.). Human factors in transport research (Vol. 2, pp.
311-319). Academic Press, London.
Canudas-de-Wit, C., Bechart, H., Claeys, X., Dolcini, P. and Martinez, J.1. (2005). Fun-to-
drive by feedback. European Journal of Control, 11(4/5),353-383.
Connolly, T. and Aberg, L. (1993). Some contagion models of speeding. Accident Analysis
and Prevention, 25(1), 57-66.
Damasio, A.R. (1994). Descartes' error: Emotion, reason, and the human brain. G. P.
Putnam's Sons, New York.
De Vos, A.P., Theeuwes, J., Hoekstra, W. and Coemet, M.1. (1997). Behavioral aspects of
automatic vehicle guidance. Relationship between headway and driver comfort. Trans-
portation Research Record, 1573, 17-22.
Donges, E. (1978). A two-level model of driver steering behavior. Human Factors, 20,
691-707.
Evans, L. (1991). Traffic safety and the driver. Van Nostrand Reinhold, New York.
Fajen, B.R. (2005). Calibration, information, and control strategies for braking to avoid a
collision. Journal of Experimental Psychology - Human Perception and Performance,
31(3), 480-501.
Finucane, M.L., Alhakami, A., Slavic, P. and Johnson, S.M. (2000). The affect heuristic in
judgments of risks and benefits. Journal ofBehavioral Decision Making, 13(1), 1-17.
Fischhoff, B., Slavic, P., Lichtenstein, S., Read, S. and Combs, B. (1978). How safe is
safe enough? A psychometric study of attitudes towards technological risks and benefits.
Policy Sciences, 9, 127-152.
Fuller, R. (2000). The task-capability interface model of the driving process. Journal
Recherche-Transport-Securite (RTS), 66,47-59.
Fuller, R. (2004). Zero risk versus target risk: Can they both be right? Paper presented at
the 3rd International Conference on Traffic and Transport Psychology 2004, Symposium
on Traffic Psychology Theories. Nottingham, UK.
Fuller, R. (2005). Towards a general theory of driver behaviour. Accident Analysis and
Prevention, 37,461-472.
Gameiro da Silva, M.C. (2002). Measurements of comfort in vehicles. Measurement Science
and Technology, 13, R41-R60.
Gibson, J.1. (1979). The ecological approach to visual perception. Houghton Mifflin,
Boston.
Gibson, J.1. and Crooks, L.E. (1938). A theoretical field-analysis of automobile driving.
American Journal of Psychology, 51,453-471.
Godthelp, H. (1988). The limits of path error-neglecting in straight lane driving. Ergonomics,
31,609-619.
 11. Towards Understanding Motivational and Emotional Factors 205

Godthelp, H., Milgram, P. and Blaauw, GJ. (1984). The development of a time-related
measure to describe driving strategy. Human Factors, 26, 257-268.
Goodrich, M.A., Stirling, W.C. and Boer, E.R. (2000). Satisficing revisited. Minds and
Machines, 10(1),79-110.
Greeno, J.G. (1994). Gibson's Affordances. Psychological Review, 101(2),336-342.
Groeger, J. (2000). Understanding Driving. Psychology Press, Hove.
Haglund, M. and Aberg, L. (2000). Speed choice in relation to speed limit and influences
from other drivers. Transportation Research Part F: Traffic Psychology and Behaviour,
3(1), 39-51.
Hall, E.T. (1966). The hidden dimension. Doubleday, Garden City.
Hancock, P.A. and Caird, J.K. (1993). Experimental evaluation of a model of mental work-
load. Human Factors, 35,413-429.
Hancock, P.A. and Warm, lS. (1989). A dynamic model of stress and sustained attention.
Human Factors, 31,519-537.
Harms, L. (1991). Variation in drivers' cognitive load: Effects of driving through village
areas and rural junctions. Ergonomics, 34, 151-160.
Hart, S.G. and Staveland, L.E. (1988). Development of the NASA-TLX (Task Load Index):
Results of empirical and theoretical research. In P.A. Hancock and N. Meshkati (Eds.).
Human mental workload (pp. 139-183). Elsevier, Amsterdam.
Hollnagel, E. (2002). Time and time again. Theoretical Issues in Ergonomics Science, 3(2),
143-158.
Hollnagel, E. and Woods, D.D. (2005). Joint cognitive systems. CRC Press, Boca Raton,
FL.
Jiang, Z.Y., Streit, D.A. and El-Gindy, M. (2001). Heavy vehicle ride comfort: Literature
survey. Heavy Vehicle Systems - International Journal of Vehicle Design, 8(3/4), 258-
284.
Lawton, R., Parker, D., Stradling, S.G. and Manstead, A.S.R. (1997). Predicting road traffic
accidents: The role of social deviance and violations. British Journal ofPsychology, 88,
249-262.
Lee, D.N. (1976). A theory of visual control of braking based on information about time-
to-collision. Perception, 5, 437-459.
McKenna, F.P. (1982). The human factor in driving accidents: An overview of approaches
and problems. Ergonomics, 25, 867-877.
Naatanen, R. and Summala, H. (1974). A model for the role of motivational factors in
drivers' decision-making. Accident Analysis and Prevention, 6, 243-261.
Naatanen, R. and Summala, H. (1976). Road-user behavior and traffic accidents. North-
Holland/American Elsevier, AmsterdamlNew York.
Niittyrnaki, J. and Pursula, M. (1994). Valo-ohjattujen liittymien simulointi ja ajodynami-
ikka (Simulation ofsignalized intersections and driving dynamics) (Publication No. 81).
Helsinki University of Technology, Transportation Engineering.
Ohta, H. (1993). Individual differences in driving distance headway. In A.G. Gale et al.
(Eds.). Vision in vehicles (Vol. 4, pp. 91-100). Elsevier, Amsterdam.
Olson, P.L. and Sivak, M. (1984). Discomfort glare from automobile headlights. Journal of
the Illuminating Engineering Society, 13, 296-303.
Ouellette, lA. and Wood, W. (1998). Habit and intention in everyday life: The multiple
processes by which past behavior predicts future behavior. Psychological Bulletin, 124,
54-74.
Rasanen, M. and Summala, H. (2000). Car drivers' adjustments to cyclists at roundabouts.
Transportation Human Factors, 2, 1-17.
206 Summala

Reymond, G., Kemeny, A., Droulez, 1. and Berthoz, A. (2001). Role of lateral acceleration
in curve driving: Driver model and experiments on a real vehicle and a driving simulator.
Human Factors, 43(3), 483-495.
Robertson, L.S. and Pless, LB. (2002). For and against: Does risk homoeostasis theory have
implications for road safety. British Medical Journal, 324, 1151-1152.
Rothengatter, T. (1988). Risk and the absence of pleasure: A motivational approach to
modelling road user behaviour. Ergonomics, 31, 599-607.
Roufs, J.AJ. and Boschman, M.C. (1997). Text quality metrics for visual display units .1.
Methodological aspects. Displays, 18(1), 37-43.
Rumar, K. (1988). Collective risk but individual safety. Ergonomics, 31, 507-518.
Rumar, K., Berggrund, U., Jernberg, P. and Ytterbom, U. (1976). Driver reaction to a
technical safety measure: Studded tires. Human Factors, 18, 443-454.
Russell, J.A. and Barrett, L.F. (1999). Core affect, prototypical emotional episodes, and
other things called emotion: Dissecting the elephant. Journal of Personality and Social
Psychology, 76(5),805-819.
Simon, H. (1955). A behavioral model of rational choice. Quarterly Journal ofEconomics,
69,99-118.
Sivak, M., Flannagan, M., Ensing, M. and Simmons, CJ. (1991). Discomfort glare is task
dependent. International Journal of Vehicle Design, 12, 152-159.
Slovic, P., Finucane, M.L., Peters, E. and MacGregor, D.G. (2004). Risk as analysis and
risk as feelings: Some thoughts about affect, reason, risk, and rationality. Risk Analysis,
24(2), 311-322.
Smeed, R.J. (1949). Some statistical aspects of road safety research. Journal of the Royal
Statistical Society, Series A (General), 112, 1-34.
Summala, H. (1985). Modeling driver task: A pessimistic prediction? In L. Evans and R.C.
Schwing (Eds.). Human behavior and traffic safety (pp. 43-65). Plenum, New York.
Summala, H. (1988). Risk control is not risk adjustment: The zero-risk theory of driver
behaviour and its implications. Ergonomics, 31,491-506.
Summala, H. (1996). Accident risk and driver behaviour. Safety Science, 22,103-117.
Summala, H. (1997). Hierarchical model of behavioural adaptation and traffic accidents.
In T. Rothengatter and E. Carbonell Vaya (Eds.). Traffic and transport psychology (pp.
41-52). Pergamon, Amsterdam.
Summala, H. (2000). Brake reaction times and driver behavior analysis. Transportation
Human Factors, 2, 217-226.
Summala, H. (2005). Traffic psychology theories: Towards understanding driving behaviour
and safety efforts. In G. Underwood (Ed.). Traffic and transport psychology (pp. 383-
394). Elsevier, Oxford.
Summala, H. and Koivisto, I. (1990). Unalerted drivers' brake reaction times: Older drivers
compensate their slower reactions by driving more slowly. In T. Benjamin (Ed.). Driving
behaviour in a social context (pp. 680-683). Paradigme, Caen.
Summala, H. and Merisalo, A. (1980). A psychophysical method for determining the effect
of studded tires on safety. Scandinavian Journal of Psychology, 21, 193-199.
Summala, H. and Rasanen, M. (2000). Top-down and bottom-up processes in driver behavior
at roundabouts and crossroads. Transportation Human Factors, 2, 29-37.
Sylwan, C. (1919). Trafiksakerheten pa gata och landsvag (Traffic safety on streets and
highways). Industritidningen i Norden (No. 50), 357-364.
Taieb-Maimon, M. and Shinar, D. (2001). Minimum and comfortable driving headways:
Reality versus perception. Human Factors, 43(1), 159-172.
 11. Towards Understanding Motivational and Emotional Factors 207

Taylor, D.H. (1964). Drivers' galvanic skin response and the risk of accident. Ergonomics,
7,253-262.
Theeuwes, 1., Alferdinck, J. and Perel, M. (2002). Relation between glare and driving
performance. Human Factors, 44(1),95-107.
Vaa, T. (2004). Developing a driver behaviour model based on emotions and feelings:
Proposing building blocks and interrelationships. Paper presented at the 3rd International
Conference on Traffic and Transport Psychology 2004, Symposium on Traffic Psychology
Theories. Nottingham, UK.
Van der Horst, R. and Wilmink, A. (1986). Drivers' decision making at signalized intersec-
tions: An optimization of the yellow timing. Traffic Engineering & Control, 27, 615-622.
Van der Horst, A.R.A. (1990). A Time-Based Analysis of Road User Behaviour in Normal
and Critical Encounters. TNO Institute for Perception, Soesterberg.
Verplanken, B., Aarts, H., van Knippenberg, A. and Moonen, A. (1998). Habit versus
planned behaviour: A field experiment. British Journal of Social Psychology, 37, 111-
128.
Warren, W.H. (1984). Perceiving affordances - Visual guidance of stair climbing. Journal
of Experimental Psychology - Human Perception and Performance, 10(5),683-703.
Warren, W.H. and Whang, S. (1987). Visual guidance of walking through Apertures -
Body-scaled information for affordances. Journal ofExperimental Psychology - Human
Perception and Performance, 13(3),371-383.
Wikman, A.-S., Laakso, M. and Summala, H. (1997). Time sharing of car drivers as a func-
tion of driving speed and task eccentricity. In The 13th Triennial Congress ofInternational
Ergonomics Association. Tampere, Finland, June 29-July 4, 1997.
Wikman, A.S., Nieminen, T. and Summala, H. (1998). Driving experience and time-sharing
during in-car tasks on roads of different width. Ergonomics, 41(3), 358-372.
Wikman, A.S. and Summala, H. (2005). Aging and time-sharing in highway driving. Op-
tometry and Vision Science, 82(8), 716-723.
Wilde, GJ.S. (1982). The theory of risk homeostasis: Implications for safety and health.
Risk Analysis, 2, 209-225.
van Winsum, W. and Heino, A. (1996). Choice of time-headway in car-following and the
role of time-to-collision information in braking. Ergonomics, 39(4), 579-592.
Williams, A.F. and O'Neill, B. (1974). On-the-road driving records of licensed race drivers.
Accident Analysis and Prevention, 6, 263-270.
Wong, Y.D. and Nicholson, A. (1992). Driver behavior at horizontal curves - risk compen-
sation and the margin of safety. Accident Analysis and Prevention, 24(4), 425-436.
12
Modelling Driver Behaviour
on Basis of Emotions and Feelings:
Intelligent Transport Systems and
Behavioural Adaptations
TRULS VAA

12.1 Introduction

Intelligent transport system (ITS) is a generic concept, which covers a wide range of
systems. In this context the concept is applied on automotive systems and comprises
systems generally defined as (advanced) driver assistance systems (ADASIDAS),
in-vehicle information systems (IVIS) and roadside telematics (RT). The present
text focuses on anti-locking brake systems (ABS), which is used as an illustrative
example of an ITS, mainly because evaluation studies have shown unintended ef-
fects that call for explanations. ABS, which aims to maintain the steering capacity
during (heavy) braking by preventing the wheels from locking, is considered a
driver assistance system (DAS). ABS has become an increasingly standard equip-
ment of new car makes and has been around for more than 20 years. Several studies
have evaluated the effects of ABS on behaviour and accidents and the system is a
case of special interest for several reasons: One is the demonstration of risk com-
pensation associated with ABS and other reasons are contra-intuitive and even
detrimental effects on traffic safety. With ABS as an illustrative example, several
key issues can be discussed when considering ITS in a more generic sense. Fur-
ther, to better understand and predict effects of ITS, a theoretical driver behaviour
model based on emotions and feelings is presented. Behavioural adaptation and
risk compensation are regarded as core problems, which have to be addressed
in terms of traffic safety. One of the very aims of the proposed driver behaviour
model is to explain and predict risk compensation that might be associated with a
given ITS.

12.2 Defining Motivation

Motives can be defined as 'factors, which give behaviour energy and direction'
(Atkinson et aI., 1996). Then, motives are factors, which initiate and govern be-
haviour. The energy component of the definition shows that a motive basically also
is a drive. The direction component implicitly presupposes repulsion or attraction,
that is, a movement away from something or attraction to something, which again

208
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 209

implies that the repulsive or attractive object has to be loaded with some emotional
quality. Otherwise, there would be no energy and no direction (Overskeid, 2000).
A neutral object is neither repulsive nor attractive. Hence, the emotional dimension
of motives is then a core aspect of motivation.
A second feature of the definition is that it does not say anything about cognition,
that is, whether motives are rooted in consciousness or in the unconscious. It may
seem self-evident as the role of the unconscious in psychoanalysis and psychody-
namic theory is well known, but I nevertheless regard this issue as a significant
point, for two reasons:

1. the role of the unconscious has seldom been made explicit in prevailing driver
behaviour models, and
2. the phenomenon of risk compensation cannot reach any satisfactory explana-
tion, without including and addressing unconscious processes.

A central concept in biological psychology is the principle of homeostasis, that

is, when basic physiological needs are regulated by a homeostatic mechanism as
with body temperature, thirst, hunger, sleep, etc. When deviances become larger
than the body can regulate internally, the organism will be motivated for a given
(external) behaviour resulting in a restored (internal) homeostatic state. Common
to psychoanalytic and homeostatic theories on motivation is the way they look upon
behaviour as initiated by states of tension in the organism, which in turn leads to
specific, purposeful acts when the state of tension exceeds a given threshold. The
behavioural goal is then to satisfy the need in order to reduce tension.

12.3 Motivational Aspects in Driver Behaviour Models

One can hardly say that the task of modelling driver behaviour has reached con-
sensus. There is no breakthrough or 'great unified theory' within the field of traffic
safety research regarding modelling of driver behaviour (Vaa, 200 1a). Models ad-
dress diverging aspects, several 'favourite' issues and/or concepts are pursued,
discussions and disagreement prevail. The listing below includes some of the most
predominant theories and models that have been applied to explain and predict
driver behaviour and which have motivational aspects as a key factor. The history
of models starts in 1938 when Gibson and Crooks' presented their theoretical field-
analysis of automobile driving (motivational aspects or motivational processes are
stated in short for each of the models):

'Field of safe travel' (Gibson and Crooks, 1938)

'Driving as a self-paced task governed by tension/anxiety' (Taylor, 1964)
'Zero-risk model' (Naatanen and Summala, 1974)
'Target level of risk' (Wilde, 1982)
'The threat-avoidance model' (Fuller, 1984)
'Theory of planned behaviour' (Aijzen, 1985)
'The role of pleasure' (Rothengatter, 1988)
210 Vaa

'Sensation seeking' (Zuckerman, 1994)

'Task difficulty' (Fuller, 2000).

The above listing is by no means a complete list of theories and driver behaviour
models, the purpose of presenting it is to focus on the main motivational aspects
which have been proposed, applied and discussed within traffic safety research.
It can be argued that a common denominator for most of the models is emo-
tion: 'Safe travel' (Gibson and Crooks, 1938), 'tension/anxiety' (Taylor, 1964),
'zero risk' (Naatanen and Summala, 1974), 'target risk' (Wilde, 1982), 'threat-
avoidance' (Fuller, 1984), 'pleasure' (Rothengatter, 1988) and 'difficulty' (Fuller,
2000). Wilde's RHT, which has been heavily debated since it was launched, repre-
sents something different, because the target level of risk is understood and defined
as a number, that is, as a number> 0 (Vaa, 2001a), not as an emotion or a feeling.
On the other hand, Wilde's RHT is inescapable regarding the discussion of models
because of its reliance on central concepts as homeostasis and risk compensation.
There is a need, however, for a reorientation of RHT, because the debate somehow
has resulted in a deadlock for the development of driver behaviour models. Such
a reorientation is also strongly needed because risk compensation is still not fully
understood and accounted for in a satisfactory way.
One may get the impression that models have been too focused on cognitive
aspects as determinants of driver behaviour, as in theory of reasoned action (TRA)
(Aijzen and Fishbein, 1980) and theory of planned behaviour (TPB) (Aijzen, 1985).
One could even say that the focus on cognitive models has been predominant to
such an extent that the role of the unconscious has more or less been neglected.
Extending this assertion, I would argue that there is no common understanding of
driver behaviour that is based on recent achievements in cognitive psychology and
neurobiology. In fact, Taylor's early work of 1964 may be more in line with recent
achievements in neurobiology than any other of the models listed above (Damasio,
1994; Bechara et al., 1997). No deep understanding of risk compensation will
emerge unless recent developments in cognitive psychology and neurobiology are
integrated in the modelling of driver behaviour (Vaa, 2001a).
The hypothesis which is proposed is that the role of the unconscious is significant
as a motivating force also when it comes to driver behaviour. Hence, there is a need
for a deeper and more considerate elaboration of the role of the unconscious. This
is made a predominant point in the present discussion, which aims at a deeper
understanding of the phenomenon of risk compensation.

12.4 Behavioural Adaptation and Risk Compensation

All living organisms have to adapt to their environments in order to survive. We
adapt continuosly to changing characteristics of the environments. Road traffic
is a specific case of environment, which calls for continous adaptations to reach
our destinations without being involved in accidents. This is self-evident. How-
ever, behavioural adaptations have certain meanings when it comes to road traffic,
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 211

especially for drivers, which call for specific definitions. Behavioural adaptation
and risk compensation are concepts which sometimes are used interchangeably,
but here a distinction between them is proposed. Behavioural adaptation is nat-
urally the widely used generic concept and this meaning remains. However, as
our perspective is road traffic and especially driver behaviour, I will limit be-
havioural adaptation to strategic decisions, that is, to conscious decision mak-
ing. Strategic decision making may take place outside as well as inside the road
traffic system. One example is driving in darkness, when the number of elderly
drivers and women (all ages), increases as a function of road lighting on a given
stretch of road, because these specific driver groups feel more secure when a road
is lit up by road lighting (Assum et al., 1999). A second example is when you
decide to drive faster because you are out of time or you decide to take an al-
ternative route because you are stuck in traffic, these are, likewise, also strategic
decisions.
Risk compensation also represents behavioural adaptation, but I regard it as a
special case of adaptation, that is, adaptations which predominantly occur without
involving consciousness. Hence, this concept is used only for decisions made on
an unconscious level, as when the driving speeds are increased (Aschenbrenner et
al., 1987) or when time headways are reduced (Sagberg et al., 1997), for drivers
driving cars equipped with ABS compared to a control group of drivers with cars
without ABS. This distinction is deliberately made because these kinds of decisions
origin in bodily reactions, that is the hypothesis, which drivers do not necessarily
experience at a conscious level. This type of process is named risk compensation,
because, when a given, supposed risk-reducing measure is introduced in the road
traffic system (here: Cars with ABS), the risk-reducing effects, which are expected,
are compensated by certain behaviour changes, most predominantly by increased
driving speeds or by changes of levels of attention (Elvik and Vaa, 2004).
It should be added that the 'hierarchical' categorisation strategic-tactic-
operational introduced by Michon (1985) is not adopted here, behaviour is rather
seen as belonging to a continuum ranging from highly conscious to completely
unconscious as end points, i.e. not as separate and distinct categories in itself (Vaa,
2003a). The degree of conscious/unconscious information processing and deci-
sion making is then understood as going back and forth along the continuum, thus
illustrating the dynamics and integration of cognitive processes, bodily reactions
and emotions and feelings.

12.5 Wilde's Risk Homeostasis Theory (RHT)

Why considering Wilde's theory of risk homeostasis in more depth than other driver
behaviour models? There are several reasons for this: (1) It has been central for
years and it has been at the core of heavy debates since it was first published, (2) it
addresses risk compensation which certainly exists and remains as an unsolved
and not fully understood problem in traffic safety work, (3) in its radical form
it represents a deadlock theoretically speaking, as it is not suitable for testing,
212 Vaa

a
I. Benefits expected from risky behaviours (.)
2.Costs expected from cautious behaviours (+)
3.Benefi1s expected from cautious behaviours (-)
4. Costs expected from risky behaviours (-)

c
Individual estimates 01 the
intrinsic effect of 0 new t - - -_ _. .
non - mofivationo1 occident
countermeasure
d
Individual levels of
perceived. rood-accident
risk

FIGURE 12.1. Wilde's model of risk homeostasis (after Wilde, 1982, 1988).

and finally (4) RHT is inescapable, it addresses a core problem in driver decision
making, there is an essence in it, which I will try to extract by contrasting Naatanen
and Summala's 'zero-risk' -model with Wilde's RHT.
While Naatanen and Summala (1974) postulate that drivers try to avoid risk by
regulating their behaviour according to a perception of zero risk, Wilde postulates
the opposite by stating that drivers seek a certain risk level- 'a target risk level' -
a risk level that must be perceived as a number > 0 presumed to be defined by
a measure of exposure, that is, as number of accidents per kilometres driven, a
certain unit of time or the like (Vaa, 2001a). This target risk level varies between
drivers, it seems partly to have idiosyncratic origins, partly to be a regulator in
a homeostatic system: When the driver is confronted with certain changes in the
road environment, he or she will meet these changes with adaptations that secures
that the level of target risk is sustained. Wilde postulates further that the target level
of risk can be increased when expected benefits from risky behaviour or expected
costs from cautious behaviour, increases. And finally, it can be reduced when
expected benefit from cautious behaviour or expected costs from risky behaviour,
increases.
Wilde's RHT model contains one explicit element called a comparator. This
is a place, a function, a process where three input factors are put together and
compared: b, c and d (Fig. 12.1) resulting in one output factor e, where:
b == Individual levels of target risk
c == Individual estimate of the intrinsic effect of a new, non-motivational road
safety measure
d == Individual levels of perceived road-accident risk
e == Desired adaptation satisfying the formula: b - c - d == 0
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 213

According to Wilde, the three input factors are 'weighed together' in the com-
parator. This is a bound weighing: RHT predicts that the end result should be zero.
All three input factors must be comprehended as numbers where the values of b,
c and d have the property that
b-c-d==O (I)

Translated into words, it means that the output from the comparator, the result
of the weighing procedure, must be chosen in such a way that the property (I) is
fulfilled. Translated to behaviour, it means that the output factor must be regarded
as the desired adaptation of the individual driver, which is such that the risk
homeostasis is sustained on an individual basis.
A problem with RHT, which in my opinion must be read according to the
above, is that all input factors and the predicted output factor, are comprehended
as numbers. And, as such, they should be confirmed to exist by individual drivers.
Are they? I have never seen any such numbers or calculating procedures being
confirmed by drivers and I have no such numbers or comparing in myself. If they
exist, the prediction must take place unconsciously. The prediction would hence
be impossible to test, as the entities are impossible to observe and impossible to
measure.
Looking more closely at the factor d, individual levels of perceived road-accident
risk, d seem to be the time-lagged feedback product (h) of a preceding aggregate
accident loss in the population (g). I guess we all have a perception of an 'aggregate
accident loss', but it is not very accurate. It may be uttered in terms of 'speeds
are increasing', 'traffic is getting worse' or 'traffic seem to improve now', but
such terms are very crude and not based on any kind of calculation or unbiased
knowledge of the situation. We may remember certain accidents, because of their
magnitude or some other characteristic, but we may have forgot all the ones with
minor injuries or damage. Tversky and Kahneman (1974) show that people put
much weight into their own experiences. The properties of one small sample may,
independent of its size, be considered as being representative of a much larger
population. Tversky and Kahneman are, however, not referred to in Wilde (1982,
1988). There seem to be no trace in the literature of the asserted weighing procedure
between b, c and d. that is, between target level of risk (b), individual estimates
of effects (c) and individual levels of perceived risk (d) constituting the desired
adaptations b - c - d == O. Not even on a strategic level of driver behaviour, that
is, thinking supposed to take place in a highly conscious and rational manner, does
it seem possible to observe these kinds of cognitive operations.
As a conclusion, a target level of risk cannot be a number, a thought or an
imagination that I bring with me consciously and which I put into some weighing
procedure when I decide what speed I should choose or what kind of acts I should
perform as was it a constant, predominant thought or imagination in the dynamics
of my thinking. And that is exactly my critique against Wilde: The RHT model does
not grasp or mimic the varied dynamics of thinking and feeling, 'the streams of
consciousness', the fluctuations of automated states mixed with thoughts coming
and going so characteristic of everyday driving. The RHT model somehow assumes
214 Vaa

Perceived risk V

X Speed

FIGURE 12.2. Hypothetical distribution of perceived risk according to driving speed.

a powerful, hidden, unconscious force that forces us to act in such a way that the
target level of risk is sustained individually for everyone as well as for everybody
else. Such a powerful force somehow resembles the cosmological anti-gravity
force, 'dark matter', 'dark energy' or whatever: The force is there, it makes the
universe accelerate in its expansion, but we cannot observe it (Vaa, 2001a).

12.5.1 Target Risk or Target Feeling?

So what is in Wilde's RHT? Let me start with the 'zero-risk' model of Naatanen and
Summala (1974). An experience of 'zero risk' will be fulfilled at several driving
speeds, in fact any number in the interval [0, Xl] may satisfy the experience of a
zero risk (Fig. 12.2):
Why stop at Xl as the chosen speed? Why not choose any speed < Xl? Why ex-
actly Xl? In my view, Naatanen and Summala do not answer this question directly.
But let us see what happens if we loosen Wilde's tight and rigid presupposition
of regarding the target level of risk as a certain number or level that the individ-
ual driver seeks to achieve or sustain. Let us suppose that this target is of another
nature. Let us instead presume that drivers are searching for a certain feeling , a cer-
tain way of driving that suits him or her well, that gives the driver' a best feeling' .
This is exactly what I would characterise as Wilde's contribution by his RHT, the
introduction of the target. However, the target should not be regarded as a number,
but as a certain kind of feeling. Furthermore, a fulfilment of a 'zero-risk' is not
sufficient, there has to be another dimension added to it as well. A dimension, an
experience, a feeling, that is achieved at the exact speed of Xl, but not at speeds <Xl
(Vaa, 2001a). This is then my hypothesis, which could be stated more generally:
Assertion: In addition to avoid accidents, drivers seek a certain 'target feeling' . This feeling
is not the same in all drivers, all drives have a unique target feeling, which is not neces-
sarily experienced consciously. Targets can be defined and characterised by an emotional
dimension, either positively or negatively.

Candidates for target feelings are as follows:

Tolerable tension/anxiety
Vigilant, attentive, highly aware ('arousal')
Sensation (seeking)
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 215

Joy/pleasure
Relaxed, secure
Threat-avoidance
Avoiding/reducing difficulties
Compliance/rule-based driving: Avoid violations, no errors, 'always behave cor-
rectly'
Non-compliance

It is worth noting that several of the driver behaviour models mentioned previ-
ously do have inherent emotional aspects, but, with the exception of Taylor (1964)
and Naatanen and Summala (1974) none of the models view emotions and feel-
ings as a governing principle in a general way i.e. only in a specific way, they
somehow isolate single feeling dimensions as their key variable as with threat-
avoidance (Fuller, 1984), joy/pleasure (Rothengatter, 1988) and sensation seeking
(Zuckerman, 1994). Three other feeling dimensions as motivating forces are also
suggested: 'Arousal' (being vigilant, attentive), compliance/rule-based driving and
non-compliance. The idea is to make the picture complete by not singling out one
specific feeling, but rather try to grasp a more complete variety of feelings that
may govern driver behaviour. Drivers are different, some feelings might be more
predominant than others. The predominance of certain feelings are likely to be
associated with personality traits, as suggested by Ulleberg (2002).
Not all drivers enjoy driving, so the 'best feeling' that can be achieved or sus-
tained may be negatively defined, as an optimal choice where unpleasantness, dif-
ficulties, etc., are reduced to their minimum in any given situation. It is proposed
that such choices are at least two-dimensional. Avoiding accidents, 'zero risk', is
not the full answer. A certain emotional experience has to be added. Car driving
is characterised by constantly solving problems, problems that involve thinking,
choosing and deciding between different alternatives. All alternatives, scenarios,
acts, can be characterised by an outcome that has an emotional dimension attached
to it. In fact, that emotional dimension is the very variable that enables drivers or
any other in any other situation, to evaluate and choose between alternatives. If
there is no feeling, there is no possibility for evaluating the outcomes (Damasio,
1994; Overskeid, 2000). There is no such thing as thinking and reasoning without
an emotional dimension.

12.6 Effects of ABS: An Illustrative Example of ITS

A German field study addressing the effects of antilocking brake systems (ABS)
on behaviour and accidents has become a classic study of risk compensation
(Aschenbrenner et al., 1987). Aschenbrenner et al. applied Wilde's risk homeosta-
sis theory (RHT) to predict that risk compensation might be seen among drivers
using cars equipped with ABS. As a conseqeunce, the suggested safety potential
inherent in ABS, might be reduced or even cancelled out. To test this hypothesis
Aschenbrenner et al. designed an experiment with a taxi company in Munich. The
216 Vaa

taxi company had some of their cars equipped with ABS, while others were with-
out ABS. The two groups were similar; the only difference was the fitting with and
without ABS. The drivers were randomly assigned to the groups and they were
all told, which kind of brakes their taxi had. Driver behaviour was recorded by
observers camouflaged as passengers. They all asked the taxi drivers to drive ex-
actly the same trip. Data from a total of 113 trips were recorded, evenly distributed
between taxis with and without ABS.
Driver behaviour data were collected for 18 variables. Of these, statistically
significant differences were observed for four variables (Aschenbrenner et aI.,
1987). These were as follows:

Drivers of taxis equipped with ABS were more often outside their lane than
drivers of taxis without ABS.
Drivers with ABS 'cut corners' more often than drivers without ABS.
Drivers with ABS predicted the traffic ahead to a lesser degree than drivers
without ABS.
Drivers with ABS were more often involved in conflicts with other road users
than drivers without ABS.
Driving speeds were measured at four sites along the fixed route. By one of
these, in a 60 km/h speed zone, driving speeds were measured. The driving speeds
of ABS drivers were significantly higher than among drivers without ABS.
Accidents were also recorded and analysed. The number of accidents was con-
trolled for mileage and also for seasonal variations. It turned out that cars with
ABS were involved in as many accidents as cars without ABS. Aschenbrenner
et al. conclude that driving behaviour of the ABS-taxis has been less cautious as
no effect of ABS on the number of accidents was recorded (Aschenbrenner et aI.,
1987).
What happens to drivers who drive vehicles equipped with ABS? What lie
behind the behavioural differences? Is it a matter of thinking differently than
drivers of vehicles without ABS? Do they feel differently? Do the differences in
driver behaviour have their origin in conscious processes, in unconscious processes
or both? Is it to be explained by vehicle characteristics, driver characteristics or
both? Aschenbrenner et aI. The study of is not the only one confirming behavioural
differences between drivers of ABS-vehicles and drivers of vehicles without ABS.
Sagberg et aI. (1997) found that taxis with ABS had significantly shorter headways
than taxis without ABS, but they found no relationships with driving speeds,
possibly because dense traffic at the observation site may have prevented drivers
from driving at their preferred speeds. No other behavioural differences for drivers
with cars with and without airbags were found.
Broughton and Baugha (2002) found in a postal survey that ABS does have the
potential of reducing the number of accidents, but also that many drivers have little
or no knowledge of ABS and its effects. They found an overall accident reduction
tendency of 3% (insignificant at confidence level of 90%), a tendency of increased
number of accidents by 10% (insignificant) among men aged 56+, a 16% accident
reduction among men aged 17-55 (significant) and a tendency among women of
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 217

TABLE 12.1. Effects of ABS on accidents. Percentage change in the number of

accidents. Results from meta-analysis (from Elvik and Vaa, 2004).
Percentage change in the number of accidents

Accident types that Best

Level of injury are affected estimate 95% CI

ABS - Brakes on personal cars

All vehicles All -3.5 (-44; -2.6)
Injury accidents All -5 (-8; -2)
Fatal accidents All +6 (+1; +12)
Effects on specific types of accident
Unspecified (All) Overturning accidents +22 (+11; +34)
Unspecified (All) Single ace. without overturning +15 (+9; +22)
Unspecified (All) Intersection accidents -2 (-5,+1)
Unspecified (All) Rear-end collisions -1 (-5; +3)
Unspecified (All) Collision with fixed objects +14 (+11; +18)
Unspecified (All) Collision with turning vehicles -8 (-14; -1)
Unspecified (All) Pedestrians/cyclists/animals -27 (-40; -12)

The meta-analysis is based on the following evaluation studies: Aschenbrenner et al. (1987),
Kahane (1993 and 1994), HLDI (1995), Hertz et al. (1995A and 1995B), and Evans and
Gerrish (1996).

increased number of accidents by 18% (insignificant). One reason why there is a

tendency of an increased number of accidents among women as well as among
men aged 56+, may be that some drivers have little or no knowledge of ABS and
its effects. One hypothesis is that the way in which some women and older men
use the ABS may tend to increase their risk of accidents (Broughton and Baugha,
2002).
In a comprehensive meta-analysis of accident studies on ABS, it was found that
the overall effect of ABS for personal cars was a marginal, although statistically
significant accident reduction of 3.5% (Elvik and Vaa, 2004). However, the effect
on fatal accidents went in the opposite direction as they were increased by 6%,
also statistically significant (Table 12.1).
Even more compelling are the effects on different accident types. On the one
hand, ABS seems to reduce collisions with turning vehicles and accidents with
pedestrians, cyclists and animals. A reduction of accidents with moving objects
is what would be expected, as an effect of ABS is the ability to maintain steering
capability during braking. On the other hand, however, the number of overturning
accidents, single accidents without overturning and collision with fixed objects,
are all accident types that were significantly increased. The effects on intersection
accidents and rear-end collisions, were practically zero, no changes in the number
of accidents regarding these accident types were documented (Elvik and Vaa,
2004).
ABS also reduces stop lengths on most road surfaces. Personal cars with ABS
on all wheels have considerably shorter stop lengths on wet surfaces than personal
cars without ABS (Robinson and Duffin, 1993). ABS may reduce stop lengths
218 Vaa

by 20% at speeds over 80 kmh and with maintenance of stability. But doubts
have been raised that ABS may have failed or lost its effect in critical situations.
American traffic police have addressed this question (Brandt, 1994). The situation
they wanted to investigate would probably be rare for the common driver, but
more prevalent for police patrols during an alarm, chasing a criminal or the like.
Suspicions arose among traffic police forces to the extent that they decided to inves-
tigate and measure stop lengths in critical avoidance manoeuvres. The suspicions
were confirmed, stop lengths did increase when police drivers were braking in
avoidance manoeuvres. The explanation to this phenomenon is uncertain, but con-
sider the difference between braking with ABS and with ordinary brakes: Heavy
braking with ordinary brakes in a critical situation would lock the wheels; that is,
all friction will be used to reduce speed to the disadvantage of loosing steering
control, while the braking forces with ABS would be split between reducing speed
and maintaing steering capacity. In other words, some of the friction forces be-
tween wheel and road surface is utilised for steering, resulting in less friction for
reducing the speed of the car (Brandt, 1994).

12.7 Issues Raised by the Example of ABS:

The Relevance for ITS
Why describe a system like ABS to such detail? Is it of relevance for ITS in a more
general sense? Indeed it is. Several issues can be extracted from this special case
of ABS:
1. Drivers with ABS cars seem to behave differently than drivers in cars without
ABS. ABS seems to affect speed choice, headways, lateral position and conflicts
with other road users. ABS seems 'to do something' with drivers, it affects the
driver in certain ways. What is this 'something', what exactly does ABS do with
the drivers?
2. The findings of Broughton and Baugha indicate that drivers' knowledge of the
system or rather lack of knowledge affects the way drivers understand and use
ABS. Anecdotal evidence suggests that some drivers, when imposing all their
powers on the brake pedal in an emergency, have experienced that the pedal 'kicks
back', the brake pedal 'shakes'. This kickback is a property of the ABS when
sensors shall prevent the wheels from locking, but some drivers may experience
this kickback from the pedal as 'something is wrong' (with the brakes or other),
which in turn may make drivers lift their pressure off the pedal, then reducing the
braking powers and the result may be an accident.
3. The third issue to be discussed is the Brandt hypothesis; that is, that stop lengths
may be increased because steering in critical avoidance manoeuvres may lead to
less friction for reducing the speed of the car. This could then be a case of which
the engineers who developed the ABS technology in the first place, did not foresee
every possible outcome of the effect of the system, which in tum suggests that the
initial risk analysis of the system has been insufficient.
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 219

With ABS as an illustrative example, these three issues could be of relevance in

a more basic and generic sense, that is, of relevance for ITS in more general terms
as five hypotheses can be extracted from the issues 1-3 above:

(a) The suggested effects of ITS may be counteracted and compensated by be-
havioural changes among drives. If so: Why do (some) drivers change their
behaviours as a function of a given ITS?
(b) What kind of properties elicit behaviour changes among drivers?
(c) Improper or insufficient knowledge of a given system may lead to a reduction
of the potential effect of ITS or even to detrimental effects.
(d) The scenarios incorporated in the risk analysis preceding the development of
a given ITS may be incomplete; that is, some outcomes of an ITS may be
unforeseen. Given the large variation of driving situations, driving tasks and
drivers, (some) drivers will hit the occasions where the effects are reduced,
unforeseen and/or detrimental.
(e) Technology can be defined, at least in some cases, as an extension of the
human organism, which can bring humans to situations in which the organism
is poorly suited for mastering. In principle, any ITS could have the potential
of 'transporting' a driver to situations which are difficult to master.

12.8 Adaptation - Mismatch Between Technology

and Human Capability

Human beings try to adapt to whatever environment he or she is exposed to.

Adaptation is a necessary prerequisite for survival; it is an integral part of the
survival mechanism of the human organism. Survival is regarded as the most ba-
sic motive of human beings (Damasio, 1994). Piaget has at times simply defined
intelligence as the ability to adapt (Hoff, 2002). A driver will adapt to what-
ever car he or she wishes to use. A car represents a technology with a potential
of transporting a driver to situations he or she would not have been exposed to
without using the car. A car somehow extends the organism, the car has the po-
tential of eliciting propensities in humans that may otherwise have been hidden,
repressed, not seen, unless he or she had been 'extended by the car'. Driver ag-
gression and eliciting hostility towards other road users is one example; the feeling
of status is another, the transition from powerlessness to status/power is a third.
Some drivers are even tempted to frighten people: Varhelyi (1996) has shown that
some one in 6 drivers accelerates when they see a pedestrian entering a pedestrian
crossing. There are numerous cases of drivers using the car as weapon and as a
means to injure or even kill people (Vaa, 2000). As humans beings we carry abil-
ities and propensities developed by evolution processes over a time span of some
2,000,000 years or so, propensities are brought into situations and environments
by means of technologies, which have not been part of the evolution history of
the human organism. The ability to monitor risk for avoiding accidents in traf-
fic is the very one our forefathers used to spot and escape predators. The risk of
220 Vaa

being killed by a predator some 30,000 years ago has been estimated to have been
1 in 10.
It follows that the inherent propensities and the behavioural repertoire of the
human organism may have limitations regarding the ability to cope with situations
created and provided by today's technology, which also means that we might be
misguided, misled and not adequately warned, when we enter situations brought
to us by technology, because the ability to monitor and judge risk is not adapted to
situations into which technology has 'transported' us. The expansion of the window
of opportunities is not fully accompanied with the tools the organism needs for
coping adequately with the dangers, which are provided by the technology of the
car. Given the enormous death tolls and personal injuries, the car is probably the
most unprecedented example of the mismatch and maladjustment between humans
and technology of any time.
Who is responsible for the mismatch between man and technology? Should we
demand that human beings, by their ability to think, of being conscious of what
he or she is doing, should detect, stop and refrain from situations, which he or she
is at danger? Should we demand that the human organism, with its inherent, but
limited ability to monitor and assess risk, should detect any danger in situations
created by the technological expansion of the window of opportunities?

12.9 ITS Technology May Enhance As Well As Reduce

the Window of Opportunities
It is obvious that a car 'do something' with drivers in terms of compensation
mechanisms by bringing the driver into situations, which are difficult to handle.
There is a difference between a VW 1200 1974-model which accelerate from 0 to
100 kmlh in some 30 s and a 2004-model Mercedes 200 SLK Kompressor which
accelerates from 0 to 100 km/h in some 7.5 s. The Mercedes sports car can realise
potentials of overtaking and speeding behaviours that are impossible to realise in
a 1974 VW. It follows that the basic entity to be studied should not be the driver as
such, but rather 'the-driver-in-a-car', because a car represents different limitations
as well as potentials of behavioural opportunities. In a generic way, IT-systems
do have potentials of counteracting or limiting dangerous driver behaviour by
improving risk monitoring in situations where the human organism is at enhanced
risk either because of the organism's limited ability to assess risk or by preventing
drivers from entering those dangerous situations, but the opposite is also an option,
as shown with ABS.
Electronic stability control (ESC) is a relatively new system, which is installed
in cars to prevent the vehicle from spinning, especially on slippery surfaces. It
seems to have very promising effects on accidents (Dang, 2004; Farmer, 2004;
Lie et al., 2004). Accident reductions from 30% to 67% have been reported (fatal
accidents with personal cars and single accidents with SUVs, respectively (Dang,
2004)).
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 221

It is a paradox that ESC seems to have unambiguously positive effects on ac-

cidents, which, as shown, was not the case with ABS. It is seemingly a paradox,
because ESC and ABS utilises technologies which are almost identical: Both sys-
tems use sensors to monitor the speed of each of the wheels, but, in addition, ESC
cars also have instruments which measure yaw rate and lateral acceleration. If
the discrepancy between yaw rate and lateral acceleration is increasing, individ-
ual wheels are brought to brake in order to reduce the rotation of the vehicle. To
explain the difference of effects between ABS and ESC, one hypothesis is that
the effects of ESC cannot be compensated, while ABS can, that is, by increased
driving speeds and reduced headways. Another way of stating this hypothesis,
considering for example driving speeds in curves, is that ABS may 'invite' you to
higher driving speeds, while this opportunity is denied when the car is equipped
with ESC. The same driver may experience an enlarged window of opportunities
with an ABS car, but a reduction of the same window with an ESC car. It is the
equipage that matters, the 'driver-in-the-car', it is the driver behaviour potentials
that differ, because of the dependency of the technology of the car. The general
question to be asked, when considering ITS, is again whether a given ITS will
enhance or limit behavioural opportunities.
Keeping in mind the examples of ABS and ESC above, several issues are es-
sential when considering effects of ITS on driving behaviour:

How will drivers adapt to ITS?

To what extent can behavioural adaptations of ITS be predicted?
To what extent will predicted effects of ITS be counteracted and/or compensated
by behavioural changes among drives?
To what extent is a given ITS based on a model of driver behaviour?

The scope here is not to answer all these issues, but rather to present a model
of driver behaviour that may have a potential of understanding and predicting
behavioural effects of a given ITS.

12.10 Damasio and the Somatic Marker Hypothesis

A disadvantage with previous driver behaviour models is that most of them do
not include aspects of physiology and neurology. Only Taylor, by proposing
GSR-constancy as a governing principle, includes such an aspect (Taylor, 1964).
Damasio and the neurobiological perspective he elaborates in his book, Descartes'
Error: Emotion, Reason and the Human Brain (Damasio, 1994), provides a more
basic understanding of humans that may serve well as a basis for developing a
model of driver behaviour. A new aspect in the development of the present model
compared to previous driver behaviour models is its theoretical foundation on
neurobiology, where concepts as emotions, feelings and the relationship and in-
terplay between unconscious and conscious process are central. The base for what
222 Vaa

subsequently shall be labelled 'the monitor model' is three simple statements,

which all are extracted from Damasio:

Axiom: Man's deepest and most fundamental motive is survival.

Deductions: Humans must possess a specialised ability to detect and avoid dan-
gers that threatens his or her survival. Hence, humans must possess an organ that
provides the monitoring of potential threats.
Assertion: The body is the monitor.

It follows axiomatically from the assumption that man's deepest motive is sur-
vival, that the organism must have an instrument, an organ, enabling it to monitor
its surroundings and the situations in which it acts. This organ is the organism itself,
the complete body and its inherent physiology developed by evolution through the
history of man where observation and identification of dangers have been of vital
importance. The organism taken as a whole is considered as a monitor, an organ
for surveillance whose prime task is to monitor the interior, that is, the state of
the body and the exterior, that is, the environment and other actors with which the
organism interact.
Damasio postulates a relationship between internal states and external behaviour
when the human organism is exposed to certain strain and emotional stress, which
forms:

.... a set of alterations [which] defines a profile of departures from a range of average states
corresponding to a functional balance or homeostasis, within which the organism's economy
probably operates at its best, with lesser expenditure and simpler and faster adjustments.
(Damasio, 1994)

A central concept in the above citation is the functional balance. This functional
balance is defined as the target feeling, which was discussed previously. This
target feeling is a kind of state that drivers are seeking to achieve and/or maintain
while driving. The drive to achieve a functional balance is regarded as a central,
predominantly unconscious knowledge, which the organism possesses about itself
and which the organism is actively seeking to maintain or to restore.
Damasio states his model by saying that something important happens before
thinking and reasoning. If, for example, a situation seems to develop into some-
thing threatening or dangerous, a feeling of unpleasantness will enter the body,
an unpleasant 'gut feeling' may be under way. Because this emotion is knit to the
body, Damasio labels it somatic ('soma' is Greek for 'body') and marker because
the emotion marks the picture or the scenario. Damasio describes the consequence
of this somatic-marker in the following way:

[A somatic marker] ... .forces attention on the negative outcome to which a given action
may lead and functions as an automated alarm signal which says: Beware of danger ahead
if you choose the option which leads to this outcome....

. . . . The automated signal protects you against future losses, without further ado and then
allows you to choose from among fewer alternatives. (Damasio 1994, p. 173)
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 223

Damasio separates between emotion and feeling and limits the concept of emo-
tion to what goes on in the body of the organism, that is, the myriads of changes in
the state of the body that is induced autonomously in all its parts and organs when
the organism is exposed to a given external event. Damasio points out that a lot
of the changes in the body state, as changes in skin colour, body position, facial
expressions etc., are also visible to others. The etymological meaning of the word
emotion relates to the direction of the changes in body state as e-motion means
'movement out' (Vaa,2001b).
Damasio distinguishes specifically between emotions and feelings and limits
feeling to processes of consciously experience, consciously sensing, the changes
of the body and the mental states. Damasio distinguishes between several levels
and defines emotions and feeling as follows:

Primary emotions: Emotions that are innate and unconscious, corresponds to the
neurobiological apparatus of the newborn infant.
Secondary emotions: Emotions that are learnt and based on individual experi-
ences, accumulated by the individual- that is, as they develop into 'the emotions
of the adult'. Predominantly unconscious or pre-conscious.
Feelings: The process of 'feeling an emotion', the process of 'making an emo-
tion conscious', to feel and transform changes in body states into conscious
experiences.
This is, in short, what I will label as 'The Damasio model' . Damasio is explicitly
aware that his definitions are 'unorthodox' (Damasio, 1994). Personally I adopt his
definitions as I consider them as fruitful definitions that facilitate an understanding
of - say - driver behaviour and also fruitful for an elaboration of a model for driver
behaviour.
While primary emotions are exclusively sub-cortical and directed towards the
body, secondary emotions also include activation of numerous prefrontal cortices,
which means that secondary emotions, in addition to the sub-cortical responses of
primary emotions, also include cortical, but still unconscious responses activated
by the external stimuli. It is assumed that the cortical loop in prefrontal cortices
that is involved in secondary emotions may give access to schemas formed and
accumulated by the learning history of the individual and that this loop enables
the body to react without involving conscious processes. And furthermore, it is
this 'loop of secondary emotions' that enables the organism to act automatically in
behaviours that are 'over-learnt' - as often experienced by drivers in driving tasks
(Vaa,200lb).
Finally, to feel an emotion, it is necessary, but not sufficient, that neural sig-
nals from the viscera, muscles, joints, neurotransmitter nuclei, that is, all body
organs that are emotionally activated, are redirected towards the neo-cortex and
certain sub-cortical nuclei. The signals from the body back to cortex go through
endocrine and other chemical routes and reach the central nervous system via
the bloodstream. The feelings, that is, the conscious experience of body states
impinged by external stimuli, then establish an association between an external
object, say a given situation in traffic and an emotional body state. Hence, by the
224 Vaa

processes of feeling and emotion, the individual is able to evaluate, consider and
choose between alternative acts in a situation that demands action. The conscious-
ness needs a continuous update of 'here-and-now', of what the body does and what
it experiences. Feelings are then the conscious experience of what the body does -
by representations of emotional body states or, as Damasio puts it,

That process of continuous monitoring, that experience of what your body is doing while
thoughts about specific contents roll by, is the essence of what I call a feeling. (Damasio
1994,p. 145).

12.11 The Monitor Model

The introduction of a monitor is justified by the Damasio model and his assertion
that emotions and feelings are fundamental mechanisms which are involved in
the organism's perception and evaluation of dangers. The concept is also adopted
from Naatanen and Summala's 'zero-risk model' (1974). Hence, the monitor is
then both a concept and a principle, as well as a model for organising processes
that influence sensing, processing of information and decision making that will
affect factors outside the organism. Figure 12.3 presents the basic structure of the
monitor model:

The Monitor

Otherroadusers ~ ~ ~ ~

f! ~ .. ~ ~ Personality ~ ~
~ ~ ~ Somatl~ ~arklng ~ 1 traits ~ ..
eo
U)
cQ)
en

: Orienting :... l patterns ~ g ..

f.~~f.~~~~~I.~~::I;::JFj;t~~:;PI .: : J:~ ~ ;t: : : i~ .

'" : : :~:.: >.\~t~ ~: e:;.: : .:.:.:.:':':'. t ~~~~~.~~ ,.11
...:.:.. .......: .....

Act/Target feeling
'Best feeling'

FIGURE 12.3. Basic structure of the monitor model (from Vaa 2003).
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 225

The monitor is nothing less than the whole of the body, the whole organism.
The boundaries of the monitor (solid line) correspond to the boundary of the
body. The internal components are all elements and processes surrounded by the
solid line: Somatic marking, personality traits, motives, interaction patterns and
a residual of other factors. Deep motivation serves as a base and will influence
other components through personality traits and motives. Personality traits influ-
ence motives and dispose for idiosyncratic interaction patterns. The interaction
patterns of the individual driver can in tum elicit new, latent motives as a con-
sequence of other road users' responses on the initial act(s) of the driver. Feed-
back loops through characteristics of the vehicle, of the road environment and
of the interactions with other road users are indicated; that is, the figure of the
model is an excerpt of a certain, instantaneous time window, which is constantly
changing by new, upcoming information fed back to the organism by feedback-
loops.
A second central concept of the model, that is, in addition to target feeling, is
the account of feelings, a concept which is used to describe an internal, con-
scious process in which one is imagining two or more alternative actions or
'inner scenarios', as when considering what to do in a given situation. It is a
process of cognitive 'weighting' or 'cost-benefit analysis' of conscious, inter-
nal scenarios against each other (Damasio, 1994; Overskeid, 2000). The point is
that given alternatives/scenarios must have a dimension of feeling, say of being
attractive or repulsive, pleasant or unpleasant or the like, that are weighed to-
gether in order to come up with a decision of what would be the best solution
in a given situation. It is this process of weighing the alternatives together that
is proposed to result in some 'account of feelings' on which a given decision is
based.
In the monitor model (Fig. 12.3) the organism is, in principle, by adopting
Damasio's concept, always 'somatically marked' as it always will be filled with
sensory stimuli from the external world (sensory storage) and with internal stimuli
from somatic marking and the knowledge storage (a concept used here equiva-
lently with Damasio's 'secondary emotions'). Two routes or modes of information
processing and decision making are suggested.

1. One predominantly conscious: From the 'marking of the body' in a given situa-
tion through feelings to the account of feelings (if more than one alternative has to
be considered) and finally resulting in a certain targetfeeling (a 'best feeling' may
also be used as a proper label of the target of the act). The result of the decision,
that is, the act itself, is fed back to the organism, through and by the effects it might
have on the vehicle, on other road users and the road environment.
2. The other route or mode is predominantly unconscious: The organism is (uncon-
sciously) seeking a functional balance, which is achieved by restoring or maintain-
ing a target feeling. The organism will constantly seek to maintain this functional
balance by appropriate behaviours. As long as the functional balance is maintained
in an automated mode, there is no need for involving consciousness. Automated
mode may prevail as long as the functional balance is maintained.
226 Vaa

The two modes of information processing and decision making described above
are naturally stylised and simplified. In real driving, there is no deliberate deci-
sion to 'drive in automated mode', this should be looked upon as an unconscious
decision of the organism itself and understood as a decision which is adequate
because the driving environment is recognised as so simple and familiar that no
conscious appraisals seem necessary. Both modes aim at maintaining or restoring
a functional balance of the organism, which is achieved by the act that realises
the target feeling, either by the conscious route or by the automated route. Two
connections or 'bridges' between the unconscious and the conscious modes of in-
formation processing are indicated. One is represented by the orienting reflex; that
is, the ability to be oriented towards certain objects by a light, a sound or a smell,
which bridges the gap from automated mode to conscious mode and further to
account of feelings if necessary. The second connection is the bridge between the
boxes of 'functional balance' and 'account of feelings'. The latter bridge is sug-
gested for describing an upcoming situation in which the driver is confronted with
a conflict, a consideration of overtaking, of choosing between certain routes, of
changing driving speeds, etc. Simply stated, that is what consciousness or working
memory is there for, as an instrument for conscious appraisals when needed. The
organism will prefer and seek to be governed by an automated mode if possible, as
this mode is less costly, that is, the organism wants to economise with its cognitive
(mental) resources (Reason, 1990; Damasio, 1994). However, the organism does
not decide (consciously) to go to automated mode, this 'decision' should rather be
regarded as a property of the organism itself, that is, a 'decision' to be governed
by automated mode whenever possible.
The direction of the links or bridges between the unconscious and the conscious
is from the former to the latter, not the other way around. One predominant logic
of the model is to maintain or restore functional balance by seeking a target or
best feeling, also indicated by the arrow directly from 'motives' to 'functional bal-
ance' and finally to 'target feeling'. During automated behaviour, there is identity
between target feeling and functional balance (Vaa, 2003a).
It is an assertion that this unconscious quest for functional balance becomes
the steering principle in the model, which also may constitute a base for a deeper
understanding of risk compensation. This view is presented as an alternative to
Wilde's RHT and especially also to his concept 'target risk' (Wilde 1982); that
is, drivers are not seeking a certain risk level other than zero, as in Naatanen and
Summala's 'Zero-Risk Model'(1974). In conclusion, drivers are seeking a target
feeling rather than a target risk.
Finally, some limitations of the monitor model should be mentioned. Naturally,
there are a lot of states that may jeopardise risk monitoring, some might be covered
up under'deep motivation' , other in the residual of 'other factors' . What the model
considers is basically an average driver, of average age, average health and with
an average composition of personality traits, etc., that is, general states where the
monitoring of risk should operate optimally. We know, however, that a variety
of states increase risk. Drink driving is by far the most dangerous of conditions:
A blood alcohol concentration (BAC) of 0.05%-0.1 % has a relative risk of 10
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 227

compared to sober drivers (Vaa, 2003). The risk monitoring function is definitely
distorted in a drunken driver. To be affected by other substances also increase risk
levels, but not in the same magnitude as alcohol: Benzodiazepines, cannabis and
opiates all increase the risk of accidents as they have relative risks of 1.54, 1.70
and 1.83, respectively. Other examples are mental disorders, neurological disorders
and sleep apnoea with relative risks of 1.72, 1.75 and 3.71. Some, but not all, of
these states may be addressed and counteracted by adequate ITS solutions, as with
AlcoLock in the case of drink driving.

12.12 The Monitor Model and Prediction of Effects of ITS

It is remarkable that humans' inherent ability to handle risks is as effective and

safe as it is in road traffic. According to the most recent estimates of accident
risks in Norway (Bjernskau, 2003), the risk of a personal injury accident among
drivers is 0.18 accidents per million kilometres; that is, one driver must drive some
5.5 million km before he or she, on the average, would be injured in an accident.
Let us suppose that a driver drives from 18 to 83 years of age, that is, for 65 years.
The Norwegian average driving distance is ca 14,000 km per year. One driver will
then drive a total of appro x 910,000 km in 65 years. Hence, it takes 5.5/0.910 ~
nearly 400 years or a group of six drivers of which only one, on the average,
will experience one injury accident during a lifelong carrier as a car driver. Then,
individually speaking, it is quite safe to drive. Furthermore, it illustrates the relative
success of risk monitoring of drivers, which is remarkable taking into account that
the ability to monitor risk has been developed for a different time and for different
environments, than for the road traffic system.
One basic question to be asked is whether a given ITS would alter the functional
balance of the driver. Obviously, ABS has done just that, as shown by ABS drivers
driving faster and with shorter headways than drivers driving cars without ABS.
A second question would be whether a given ITS addresses aspects where the
organism's ability to monitor risk is weak or inadequate. No doubt, drivers' ability
to monitor risk is far from perfect. The risk monitoring ability of the human
organism is not an infallible machine, and it has weaknesses regarding monitoring
of dangers. Some important examples of monitor weaknesses that theoretically
could be improved by ITS would be:

Prevention of excessive speeds, especially among young, inexperienced drivers

who underestimate the dangers of high driving speeds. Cars fitted with elec-
tronic stability control (ESC) and intelligent speed adaptation (ISA) should be
beneficial especially for young drivers as inexperienced drivers are especially
susceptible to risk of accidents, because they lack 'the warnings' provided by the
schemes or secondary emotions, which accumulate in the organism as a function
of being repeatedly exposed to potentially dangerous situations.
The ability to detect speed changes of the car in front seem to have been poorly
developed in the human organism. The assertion is that no evolutionary selection
228 Vaa

on the ability 'to-detect-dangers-when-following-after-a-Ieader-in-a-queue' has

been necessary, because risk monitoring could be said, that is my hypothesis,
to be considered to be the responsibility of the leader. Hence, as this may be
classified as a real weakness of the inherent ability to monitor risk, alarm systems
that warn drivers of speed changes and/or changes of time headways should be
beneficial in terms of detecting dangers and thereby preventing accidents.
In cases of considering overtaking, the looks for other drivers in the rear view
and side view mirrors will leave the driver with a 'dead angle'. This is definitely
a weak point in drivers' monitoring ability regarding perception, information
processing and decision making. Alarming drivers attempting to overtake when
there is a moving object in the dead angle of the mirrors should likewise be
beneficial, then applying analogous reasoning as with car following.
Questions could likewise by raised regarding drivers' perception of motorcyclists
and pedestrians, that is, that the human organism in the role of a driver might have
inherent, weaknesses in detecting motorcyclists and cyclists in certain situations.
Some drivers fail to spot 2-wheeled road users on a crossing course, especially in
left turns. A Norwegian study found that MCs are overrepresented by a factor of
eight to one in collisions between a car and a motorcyclist on a crossing course
(Glad, 2001). One hypothesis predicted by the monitor model is that drivers
'look for dangers'; that is, drivers look primarily for other cars, as motorcyclists
do not represent a danger in the same meaning as other cars do. Then drivers
seem to overlook a motorcyclist, they sometimes might 'look without seeing'.
The same phenomenon is also seen in conflicts with cyclists.
Ordinary marked pedestrian crossings increase the number of accidents by some
28% (95% CI: +39; + 19) (Elvik and Vaa, 2004). Whatever cause, it must have
something to do with a perceptual weakness or inattentiveness by the driver, by
the pedestrian or by both. An ITS-solution which reliably could detect and warn
the driver of pedestrians should work in a beneficial way.
An ITS which warns drivers of crossing vehicles at junctions should have a
potential of reducing accidents, especially among elderly drivers who are over-
represented in accidents at junctions.
Sleep apnoea and fatigue increase accident risk considerably and much more
than being influenced drugs as benzodiazepines, anti-depressant, cannabis and
opiates (Vaa, 2003b). Drivers' appraisal of their ability to stay awake while driv-
ing seems very insufficient. Hence, systems that monitor and warn drivers of
attention failures and incidents of falling asleep should, in principle, be ben-
eficial in terms of traffic safety. It seems, however, to be quite difficult to de-
velop reliable algorithms that could warn drivers of falling asleep, probably
because the markers of sleepiness and fatigue vary considerably between drivers
as well as within an individual driver. Hence, with reference to the ED projects
AWAKE and SENSATION, a high frequency of false alarms might probably be
expected.

The above group of IT-system ideas would generally, as they all address the
primary task of driving, be labelled driver assistance systems (DAS) or advanced
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 229

driver assistance systems (ADAS), although the difference between these two
groups of systems is not well defined. The examples address situations and con-
flicts where drivers may have difficulties with inattention and/or in perceiving the
presence of other road users. The main objective is then to provide the driver with
devices that increase attention by warning him or her of upcoming dangers. The
general issue considered here regarding (A)DAS is the one of risk compensation,
that is, whether or not a given ITS will enable the driver to seek a better feeling
by increasing the speed, changing the time headway and/or the level of atten-
tion. Risk compensation has been documented to be present with ABS, but not
with ESC. For other ITS in this group of driver assistance systems the question
remains if or how they will influence the monitoring of risk by compensatory
mechanisms.

12.13 Summary and Conclusions

As documented previously, the average and experienced driver perform quite well
regarding risk monitoring as it must be regarded as normal to have a lifelong
career of driving without a single personal injury accident. The introduction of
new ITS in cars would then face an issue of trust, that is, whether or not the
driver would perceive a given system as reliable. Consequently, it is suggested
that a given ITS must perform better than the driver, because the driver gener-
ally will regard him or herself as competent and skilled in handling the dangers
of everyday driving. It is a matter of driver confidence and reliance of a sys-
tem. ITS must prove its relevance and its reliability. If false alarms occur too
often, it will create mistrust and subsequently the driver will probably abandon the
system.
Initially, an attempt to categorise and define different subgroups of ITS was
made. In the present discussion only (A)DAS have been considered, while IVIS
(in-vehicle-information-systems) deliberately have been left out of consideration.
The main reason for this deliberate exclusion has been that the IVIS group is
more obscure than the (A)DAS-group. While (A)DAS generally could be said
to address primary driving tasks, especially by providing information about and
warnings of dangers, IVIS may address primary as well as secondary driving tasks.
Nomadic systems, as mobile phones and portable route guidance systems belong
to the IVIS group. Hence, IVIS may represent distractions and increased workload
and thereby threats to risk monitoring, more than provisions of information about
upcoming dangers. One example is the use of mobile phones, which increases
the risk of accidents. In a study conducted by Sagberg, the relative risk of using
a mobile phone while driving was estimated to 1.72; that is, mobile phone use
increases the risk of accidents by some 72% (Sagberg, 2001). It follows that drivers
may misjudge their ability to monitor risk while operating systems which do not
address primary driving tasks. By incorporating systems that impose distractions
and increased workload, the monitoring of risk may be hampered to such an extent
that the safety of driving is jeopardised.
230 Vaa

If, or rather when, mobile phone manufacturers succeed in integrating func-

tions as Word, Excel, Internet, GPS, route guidance, MP3, TV/video and the like
and thereby reduce the need for other nomadic systems, it will provide the driver
with an enormous amount of information that is irrelevant for driving a car. Such
a scenario should be regarded as a major threat to future road safety, because it
represents possibilities of more distraction and increases in workload. So far, no
driver behaviour model seems to have adequately integrated both risk monitor-
ing and driver workload as their prime dimension. The monitor model predicts
that IT systems which address inherent weaknesses of risk monitoring in the hu-
man organism may succeed in preventing road accidents. The broader picture of
behavioural with adaptations with which drivers will meet current and future im-
plementation of ITS in cars and the road system is still to a large extent unknown.
More evaluation studies are needed. Some predictions of ITS are suggested, but
more elaborate driver behaviour models seem likewise to be needed in order to
provide better predictions of the effects of IT systems.

References
Aijzen, I. and Fishbein, M. (1980). Understanding attitudes and predicting social behaviour.
Prentice-Hall, Englewood Cliffs, NJ.
Aijzen, I. (1985). From intentions to actions: A theory of planned behaviour. In 1 Kuhl and
Beckmann (Eds.). Action control. From cognition to behaviour. Springer Verlag, Berlin,
pp. 11-40.
Aschenbrenner, K.M., Biehl, B. and Wurrn, G.W. (1987). EinfluB Der Risikokompensation
auf die Wirkung von Verkehrsicherheitsmassnahmen am Beispiel ABS. Schriftenreihe
Unfall- und Sicherheitsforschung Straj3enverkehr. Heft 63 Bergisch Gladbach, Bunde-
sanstalt fur Strassenwesen (BASt), pp. 65-70.
Assum, T., Bjernskau, T., Fosser, S. and Sagberg, F. (1999). Risk compenasation - The case
of road lighting. Accident Analysis and Prevention, 31, 545-533.
Atkinson, R.L., Atkinson, R.C., Smith, E.E., Bern, D.J. and Nolen-Hoeksema, S. (1996).
Hilgard's introduction to psychology (12th ed.). Hartcourt Brace College Publishers,
Fortworth, TH.
Bechara, A., Damasio, H., Tranel, D. and Damasio, A.R. (1997). Deciding advantageously
before knowing the advantageous strategy. Science, 275, 1293-1295.
Bjernskau, T. (2003). Risk in road traffic 2001/2002. Oslo, Institute of Transport Economics.
T0I report no 694/2003. (In Norwegian, with summary in English).
Brandt, B. (1994). ABS increases stopping distances in braking/evasive manoeuvre. Acci-
dent Reconstruction Journal, 41-42.
Broughton, J. and Baugha, C. (2002). The effectiveness of antilock braking systems in
reducing accidents in Great Britain. Accident Analysis and Prevention, 34, 347-355.
Damasio, A.R. (1994). Descartes' error: Emotion, reason and the human brain. G.P.
Putnam's and Sons, New York.
Dang, IN. (2004). Preliminary results analyzing the effectiveness of electronic stability
control (ESC) systems. Report no. DOT-HS-809-790. U.S. Department of Transportation,
Washington, DC.
Elvik, R. and Vaa, T. (2004). The handbook of road safety measures. Elsevier, Oxford.
 12. Modelling Driver Behaviour on Basis of Emotions and Feelings 231

Evans, L. and Gerrish, P.H. (1996). Antilock brakes and risk of front and rear impact in
two-vehicle crashes. Accident Analysis and Prevention, 28, 315-323.
Farmer, C. (2004). Effect ofelectronic stability control on automobile crash risk. Insurance
Institute for Highway Safety, USA Traffic Injury Prevention.
Fuller, R. (1984). A conceptualization of driving behaviour as threat avoidance. Ergonomics,
27, 1139-1155.
Fuller, F. (2000). The task-capability interface model of the driving process, Recherche
Transports Securite, 66, 47-59.
Gibson, J.1. and Crooks, L.E. (1938). A theoretical field-analysis of automobile-driving.
The American Journal of Psychology, 51(3),453-471.
Glad, A. (2001). Glare effects of high beam on motorcycles in daylight. Oslo, Insti-
tute of Transport Economics. T0I-report 521/2001. (In Norwegian with summary in
English).
Hertz, E., Hilton, 1. and Johnson, D.M. (1995A). An analysis ofthe crash experience oflight
trucks equipped with antilock braking systems. Report DOT HS 808278. U.S. Department
of Transportation, National Highway traffic Safety Administration, Washington, DC.
Hertz, E., Hilton, 1. and Johnson, D.M. (1995B). An analysis ofthe crash experience ofpas-
senger cars equipped with antilock braking systems. Report DOT HS 808279. U.S. De-
partment of Transportation, National Highway traffic Safety Administration Washington,
DC. (USA, smaller lorries and multi-purpose vehicles).
HLDI - Highway Loss Data Institute (1995). Three years on-the-road experience witn
antilock brakes. HLDI Special Report A-47. Highway Loss Data Institute Arlington, Va,
(USA, cars).
Hoff, T. (2002). Mind design: Steps into an ecology ofhuman-machine systems. Dr. Polit.
Dissertation. Trondheim, Department of Psychology and Department of Product Design
Engineering, Norwegian University of Science and Technology.
Kahane, C.1. (1993). Preliminary evaluation ofthe effectiveness ofrear-wheel antilock brake
systemsfor light trucks. Draft Report December 1993. US Department of Transportation,
National Highway Traffic Safety Administration, Washington, DC.
Kahane, C.1. (1994). Preliminary evaluation of the effectiveness of antilock brake systems
for passenger cars. Report DOT HS 808 206. US Department of Transportation, National
Highway Traffic Safety Administration, Washington, DC.
Lie, A., Tingvall, C., Krafft, M. and Kullgren, A. (2004). The effectiveness of ESP (electronic
stability programme) in reducing real life accidents. Traffic Injury Prevention, 5, 37-41.
Michon, J.A. (1985). A critical view of driver behavior models: What do we know, what
should we do? In L. Evans and R.C. Schwing (Eds.). Human behaviour and traffic safety.
Plenum Press, New York.
Naatanen, R. and Summala, H. (1974). A model for the role of motivational factors in
drivers' decision-making. Accident Analysis and Prevention, 6, 243-261.
Overskeid, G. (2000). The slave of the passions: Experiencing problems and selecting
solutions. Review of General Psychology, 4, 284-309.
Reason, J. (1990). Human error. Cambridge University Press, Cambridge.
Robinson, B.1. and Duffin, A.R. (1993). The performance and reliability of anti-lock braking
systems. Braking of road vehicles. Proceedings of the Institution of Mechanical Engi-
neers, 23-24 March 1993. Institution of Mechanical Engineers (IMechE), Birdcage Walk,
London. Published by Mechanical Engineers Publications Limited, pp. 115-126.
Rothengatter, T. (1988). Risk and the absence of pleasure: A motivational approach to
modelling road user behaviour. Ergonomics, 31, 599-607.
232 Vaa

Sagberg, F., Fosser, S. and Saetermo, LA.F. (1997). An investigation of behavioural adapta-
tion to airbags and antilock brakes among taxi drivers. Accident Analysis and Prevention,
29, 293-302.
Sagberg, F. (2001). Accident risk of car drivers during mobile telephone use. International
Journal of Vehicle Design, 26(1), 57-69.
Taylor, D.H. (1964). Driver's galvanic skin response and the risk of accidents. Ergonomics,
7,439-451.
Tversky, A. and Kahneman, D. (1974). Judgement under uncertainty: Heuristics and biases.
Science, 185, 1124-1131.
Ulleberg, P. (2002). Personality subtypes of young drivers. Relationship to risk-taking pref-
erences, accident involvement and response to a traffic safety campaign. Transportation
Research Part F, 4, 279-297.
Vaa, T. (2000). A comment on the definition of aggression and aggressive driving behaviour.
Proceedings ofthe Conference 'Road Safety on three Continents' Pretoria. South Africa,
September 2000, pp. 20-22.
Vaa, T. (2001a). Cognition and emotion in driver behaviour models: Some critical view-
points. Proceedings of the 14th ICTCT Workshop. Caserta 2001 (www.ictct.org).
Vaa, T. (2001b). Driver behaviour models and monitoring of risk: Damasio and the role of
emotions. Proceedingsfrom VTI-Conference Traffic Safety on Three Continents. Moscow
19-21 September 2001.
Vaa, T. (2003a). Survival or deviance? A Model for Driver Behaviour. Final report. Oslo,
Institute of Transport Economics. T0I-report no. 666/2003 (In Norwegian with summary
in English).
Vaa, T. (2003b). Impairments, diseases, age and their relative risks ofaccident involvement:
Resultsfrom meta-analysis. Deliverable Rl.l of EU-project IMMORTAL. Oslo, Institute
of Transport Economics, T0I report no. 690/2003.
Varhelyi, A. (1996). Dynamic speed adaptation based on information technology: A theo-
retical background. Department of Traffic Planning and Engineering, Lund Institute of
Technology, University of Lund, Bulletin 142.
Wilde, G.J.S. (1982). The theory of risk homeostasis: Implications for safety and health.
Risk Analysis, 2, 209-225.
Wilde, G.J.S. (1988). Risk homeostasis theory and traffic accidents: Propositions, deduc-
tions and discussion of dissension in recent reactions. Ergonomics, 31(4),441-468.
Zuckerman, M. (1994). Behavioural expressions and biosocial bases of sensation seeking.
Cambridge University Press, Cambridge.
 V
Modelling Risk and Errors
13
Time-Related Measures for Modelling
Risk in Driver Behaviour
RICHARD VAN DER HORST

13.1 Introduction
Accident statistics have an important general safety monitoring function and form a
basis for detecting specific traffic safety problems. However, the resulting informa-
tion is inadequate for analysing and diagnosing, defining remedial measures and
evaluating their effects. Systematic observations of driver behaviour, combined
with knowledge of human information-processing capabilities and limitations,
offer wider perspectives in understanding the causes of safety problems and mod-
elling driver behaviour in both normal and critical situations. Renewed interest
results from the need to develop, test, assess and evaluate driver support systems
in terms of drivers' behaviour, performance and acceptance.
The processes that result in near-accidents or traffic conflicts have much in
common with the processes preceding actual collisions (Hyden, 1987); only the
final outcome is different. The frequency of traffic conflicts is relatively high, and
they offer a rich information source on causal relationships since the preceding
process can be systematically observed. In this approach, traffic situations are
ranked along a continuum ranging from normal situations, via conflicts to ac-
tual collisions. A pyramidal representation of this continuum was introduced by
Hyden (1987), clearly visualising the relative rate of occurrence of the different
events (Fig. 13.1). The analysis of driver behaviour in critical encounters may
not only offer a better understanding of the processes that ultimately result in
accidents, but, perhaps even more efficient in the long run, also provide us with
knowledge on drivers' abilities of turning a critical situation into a controllable
one.
A general conceptual description of the driving task as commonly used in traf-
fic psychonomics, with time-to-line crossing (TLC) and time-to-collision (TTC)
as a measure for describing the lateral and longitudinal driving task will be used
to distinguish normal from critical behaviour. That may serve as realistic crite-
rion settings for in-car warning systems such as forward collision warning and
intersection collision avoidance warning systems.

235
236 van der Horst

undisturbed passag es

FIGUR E 13.1. The continuum of traffic events from undisturbed passages to fatal accidents
(from Hyden. 1987).

13.2 The Driving Task

In the literature the task analy sis for driving a car is well documented. A frequently
used conceptual model of the driving task con sists of three hierarchically ordered
levels, navigation, guidance and control (Allen et aI., 1971). In other publicat ions
these levels are also referred to as strategic, manoeuvring and control. Tasks at the
navigat ion level refer to the activities related to planning and exec uting a trip from
origin to destination. The need for processing information only occ urs occasion-
ally, with intervals ranging from a few minutes to hour s. The guidance level refer s
to tasks deal ing with the interaction with both environment (roadway, traffic signs ,
traffic signals) and other road users. Activity is required rather frequently with
intervals of a few seconds to a few minutes. At the control level the motion of the
vehicle is controlled in longitudinal and lateral direction. Information has to be fre-
quently processed, ranging from intermittent activities every few seconds to almost
continuous control. Alexander and Lunenfeld (1986) visuali sed the relationship
between the levels by a set of nested triangles (Fig . 13.2), hierarchically ordered
from a low level to a high level with an increasing complexity and from high to
low with an increasing urgency (primacy) . For example, a flat tire or suddenly
being confronted with a heavy wind gust will immediately interrupt activities at

FI GURE 13.2. The three hierarchical levels of

the driving task (according to Alexander and
Lunenfeld, 1986).
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 237

FIGURE 13.3. The driving task in

three dimensions (according to
>-
1:
o
INFORMATION PROCESSING 1.~'(..~e\'.f .
Theeuwes, 1993). ~ strategic .

J]
a:w manoeuvnng I - - - - - + - - - - + - - - - - r
I control
~ perception processing action
CIJ
(
~
\ _Age..based
~f\O'fJ~ ad
ru\e..b3S
s\<\\\-b3Sed

the navigation level and put all attention to the control level, since getting lost has
less severe consequences than running off the road.
At each level of the driving task successive steps of information processing; that
is, perception, processing and decision making, and action take place. Moreover,
the way a driver performs these steps strongly depends on the routine one has
developed in task performance. Rasmussen (1985) distinguishes three levels of
task performance: knowledge-based, rule-based and skill-based. The highest level
(knowledge-based) refers frequently to new situations (e.g., finding the best route
to a new destination) or situations that occur frequently in itself, but in which the
driver still has little experience. The choice of behaviour depends on interpretation
and deductive reasoning. When a situation occurs frequently, then after some time
a rule develops how to deal with that situation and recognising that situation leads
to appropriate behaviour without a 'need' to understand exactly what is going on.
Skill-based tasks are conducted automatically, incoming information automatically
results in behaviour without any cognitive control. Theeuwes (1993) introduced a
nice three-dimensional representation of the driving task as is given in Fig. 13.3.
It is obvious that also other aspects of the driver such as his intentions, attitudes,
emotions and subjective norms play a role in modelling driver behaviour. One
possible representation of this is given in Fig. 13.4 (van der Horst, 1998).

:- -
D~~~ -----------------------_. ----------------------------

Road
environment Vehicle
speed,
heading, etc.

FIGURE 13.4. A driver behaviour model according to van der Horst (1998).
238 van der Horst

13.3 Lateral Control

13.3.1 Time-to-Line Crossing (TLC)
Several traditional models on steering control are based on the assumption that the
driver acts as a path error-correcting mechanism continually allocating attention
to the steering task. Godthelp (1984) and Godthelp et al. (1985) developed the
TLC approach for describing the driving task merely as a supervisory task. It
provides a preview predictor model for time periods; for example, path errors
can be neglected. TLC can be calculated on the basis of lateral position, heading
angle, speed, and commanded steering angle, and quantifies the time needed by
the vehicle to reach either edge of the lane. Figure 13.5 shows an example from an
experiment on straight lane keeping with driver's self-paced occlusion times (Tocc ) .
Subjects made a series of runs (with a given speed of 20, 40, ... , 120 kmIh) on the
right lane of a motorway closed for other traffic with a voluntary occlusion wearing
the PLATO spectacles (Milgram and van der Horst, 1984; van der Horst, 2004).
In its normal state, the occlusion device occluded the visual field completely. On
command of the driver (by pressing the horn lever) the device switched to the open
mode for 0.55 s. The study showed a very good correlation between TLC and the
Tocc as voluntarily chosen by the subjects. Figure 13.6 indicates that drivers use a

Tocc.1 Tocc.2 Tocc.3 ToccA

yes
looking
no....., ~_..........- _.........-----
,, 0.5 S
C
o looking periods
,,
I

"~

"00 I
o
a. ~""""'-----"'----------"'-~--'---~---

~
Q)
.I

o
~I---+----+---++----+----+--+----+l----+-+---

time

FIGURE 13.5. Example of time history for time-to-line crossing (TLC) to the left and right
lane marking at the moment of looking (from Godthelp 1984).
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 239

FIGURE 13.6. Speed dependency

of median Tacc , and TLC e (from .- - 4 TLC e + Tacc
16 o . . . . . . TLC e
Godthelp et al., 1985).
...--... Tacc
14

12

:e 10
CD
.~ 8
,,,
,,
,,
6 '~ ......
.. ....
............... ....
~

4 ......
2

20 40 60 80 100 120
speed (km/h)

rather consistent internal representation of the time available to neglect path errors
dependent of speed with a Tocc at about 40% of the totally available time.
Since then, many studies illustrated the value of the TLC approach for describing
drivers' steering strategy and performance (a.o. Godthelp, 1988; Godthelp and
Kaeppler, 1988; Van Winsum et al., 2000).

13.3.2 Lateral Distance When Passing

High speeds are an important contributing factor of accidents on rural 80 km/h
roads where about 50% of all people killed in traffic in The Netherlands occurs.
An experimental project in Drenthe (one of the provinces in The Netherlands) was
aimed at the development of measures that effectively reduce speed without sig-
nificantly reducing driving comfort up to speeds of 80 km/h. First, several options
were tested in a driving simulator study (van der Horst and Hoekstra, 1994) result-
ing in the recommendation of a reduced lane width of smooth asphalt of 2.25 m, a
0.30-m central chipping strip marking, and a 0.60-m wide edge strip of 4 m long
blocks of chippings with a pacing of 4 m (see Fig. 13.7). Some concerns arose with
respect to the lateral placement of drivers due to the wide edge strip. The question
was whether drivers would choose a lateral position in the traffic lane closer to the
middle of the road with, as a consequence, a higher risk of head-on collisions. To
quantify the effects of the measures on the lateral placement of vehicles, the driving
behaviour at one experimental location was compared with that at a conventional
80 kmIh road (lane width 2.75 m and standard road delineation of a 0.10-m cen-
tre line and edge lines) by means of a quantitative analysis of video recordings
240 vander Horst

FIGURE 13.7. Speed-reducing measures as testedin Drenthe (vander Horst, 1996).

(van der Horst, 1996) (see Fig. 13.8). The analysis of cumulative distributions of
the lateral placement of free-driving vehicles (both passenger cars and trucks) at
both sites reveals that the package of speed-reducing measures does not result in
a lateral placement more to the middle of the road. On the contrary, the 0.30-m
central chipping strip marking (instead of the O.IO-mconventional central mark-
ing) even results in a lateral placement of minimally 0.10 m further away from
the centre of the road (see Fig. 13.9). Drivers mainly seem to focus on the most
nearby side of the central marking . At the moment oncoming traffic is passing ,
drivers seem to focus on the other vehicle. The mutual distance between two pass-
ing vehicles does not differ at all between the experimental and control location
(see Fig. 13.10).

13.4 Longitudinal Control

13.4.1 Time-to-Collision (ITC)
In research on traffic conflicts techniques, Hayward (1972) initiated a search for
objective measures to describe the danger of a conflict situation. He concluded
that TTC is a dominant one, being 'the time required for two vehicles to collide
if they continue at their present speed and on the same path' . TTC at the onset
of braking (ITC br ) represents the available manoeuvring space at the moment an
evasive action starts. The minimum ITC (TTCmin) as reached during the approach
process of two vehicles is taken as an indicator for the severity of an encounter;
in principle, the lower the TTCminthe higher the risk of a collision. As an exam-
ple, Fig. 13.11 indicates what happens when a car approaches a stationary object.
Usually, the concave shape of the ITC curves does not show up so clearly since
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 241

FIGURE 13.8. Quantification of lateral placement from video at a control location with
conventional markings (top) and at an experimental location in Drenthe (bottom) .

in more complex interactions between two moving road users the collision course
is often ended before point B is reached. But even then, TIC min indicates how
imminent an actual collision has been . Details of the calculation of TTC can be
found in van der Horst (1990) . He evaluated the TTC measure in normal and more
critical encounters between road users in several empirical observation studies . In
one study he analysed all encounters with a collision course between intersecting
242 van der Horst

passenger car, direction 1, free-driving

."
/ - = controlloc.
/' = exper. loc.
~ 80
c:
o
"';::;
.5
or::
eo
tn
:0
~ 40
.~

"'
's
E
~ 20

o 50 100 150 200 250 300 350

avg, lateral placement (em)

FIGURE 13.9. Cumulative distributions of the left (LW) and right (RW) wheel of free-driving
passenger cars at experimental and control location relative to the middle of the road (0 em).

passenger cars
100

- = control location
....... = experimental location
80
?ft-
c
0
'.j:i
::J
.lJ 60
~
+oJ
en
=u
Q)
>
',t:j
40
co
S
E
:J
CJ 20

0
0 50 100 150 200 250 300
mutual lateral distance when passing (em)

FIGURE 13.10. Cumulative distributions of the mutual distance between passenger cars at
the moment of passing at experimental and control locations.

road users (at a priority intersection and a general rule intersection) for a given
time period. Figure 13.12 gives the distribution of all encounters with a collision
course (van der Horst, 1991a). Only 1.6% of the 373 encounters with a collision
course displayed a TTC min of less than 1.5 s. Figure 13.13 combines these results
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 243

FIGURE 13.11. Time histories of braking by a

car approaching a stationary object; DIST, dis-
tance to object; V, velocity; ACC, acceleration
and TTC, time-to-collision based on constancy
of speed and heading angle. Point A indicates
TTC br and point B TTC min.

I
I
I
1
I
I
I
I
I
I
I
I
c
20 21 22 23 24 25 26
time Is)

t------~-----------

all encounters with a TTCmin .....-

..,._._._._e'
_ _____J 100

:....... ~
/e n= 373 80 c
/- o
<I) 15 /.

./
L- +'"
:J
~ .D
C
:J 60 L-
o UI
gOJ 10 '0
Qt
'0 40 ~
.... ~
QI
J::J ~

E 5 E
20 ~
:J
c

2 5 6

FIGURE 13.12. TTCmindistribution of all encounters displaying a collision course at two

intersections. The scale at right relates to the cumulative distribution (from van der Horst,
1991a).
244 van der Horst

. _ '.::-- ."7"::':"':'.
c
o
..,
~ 0.8 /

-
.:

-i: /
.!!! i
~ 0.6 /
.~
C
~
e
::J
0.4
u
:" --Malmo (n= 101)
0.2 - - - TrautenfeLs In- 621
- . - bicycle route (n= 1477)
....... priori ty regulation (n= 373)

2 3 4 5 6
TrCmin(S)

FIGURE 13.13. TTC min distributions of conflicts from the Malmo and Trautenfels calibration
studies, of encounters at intersections with bicycle tracks, and of the study from Fig. 13.12.

with the outcome of two calibration studies on traffic conflicts techniques under
the auspices of the ICTCT (now International Co-operation on Theories and Con-
cepts in Traffic Safety). In these studies, one in Malmo and one in Trautenfels,
the conflicts as scored by traffic conflict observation teams from eight different
countries have been analysed quantitatively in terms of TTC. Obviously, the inter-
actions scored as conflicts by conflict observer teams in the field, have much lower
TTC min values than all encounters with a collision course that occur at a yield and
general rule intersection. A bicycle route study gives a distribution somewhere in
between, mainly due to the (conservative) selection procedure as applied for that
study (van der Horst, 1990).
Based on these empirical findings, the question arises whether time measures
such as TTC are used directly by the road user as a cue for decision making. Lee
(1976) suggested that drivers are able to control braking based on TTC information
as directly available from the optic flow field. He states that information from the
optic flow field is likely to be used directly rather than information on distance
and speed explicitly. In a study on drivers' strategies of braking, it was found
that both the decision to start braking and the control of the braking process itself
may well be based on TTC information as directly available from the optic flow
field (van der Horst, 1990, 1991b). In a field experiment, subjects approaching a
stationary object (simulated rear end of a small passenger car) with a given speed,
were instructed to start braking at the latest moment they thought they could stop
in front of the object. Figure 13.14 (top) reveals that TTC br increases with speed,
but less than expected on the basis of a constant deceleration model. The effect of
braking instruction (start normal braking at the latest moment you think you can
stop safely versus hard braking at the latest moment you think you are able to stop
in front of the object) indicates that subjects are able to apply the given instruction
well, independent of approach speed. The minimum TTC as reached during the
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 245

FIGURE 13.14. TTC br (top) and

TTC min (bottom) as a function of
approach speed and braking in-
3
struction (from van der Horst and
Hogema, 1994).
:e 2

o normal braking
hard braking

30 50 70
speed (km/h)

o normal braking
3 hard braking

--~

30 50 70
speed (km/h)

approach appears to be independent of approach speed and normal or hard braking

instruction and reaches a mean value of about 1.1 s (Fig. 13.14 bottom).
In a study on the evaluation of a fog detection and signalling system, Hogema
and van der Horst (1994) analysed a huge amount of inductive loop data on an
individual vehicle-by-vehicle basis together with visibility measurements from a
nearby scatter type visibility sensor. Figure 13.15 shows the mean-free driving
speed (free driving defined as having a headway of >5 s) as a function of visibility
range on a motorway (speed limit 120 km/h) that served as a control location for the
fog warning system itself on the A 16 nearby. Two lines are added that indicate the
(maximum) initial speed possible as a function of the required stopping distance,
the left for a rather extreme case of a short reaction time (1 s) and hard braking
(acceleration -5 rn/s2 with no safety margin left). The right line corresponds to
a more moderate reaction time and deceleration and a margin of 5 m left after
246 van der Horst

Tr=2 s
150 8= ..3m/s 2

J: +
E 100 +
~
-c
Q)
Cl)
o,
en
+right lane
50 )Cleft lane

100 200 300 >1000

visibilityIbraking distance (m)

FIGURE 13.15. Mean free driving speed and visibility on the A59 motorway and initial speed
possible as a function of stopping distance given reaction time T, and required deceleration
level a.

stopping. Even in the more extreme case, the speeds of the free driving vehicles
with a visibility range between 40 and 120 m are too high to avoid a collision when
the driver is suddenly being confronted with a stationary object (e.g., a stopped
vehicle). Hogema and van der Horst (1994) computed other scenarios as well and
concluded that drivers at mean speed might be able to slow down in time for lead
vehicles that drive with a speed of at least 38 and 53 kmlh in the right and left lane,
respectively.
The results of the inductive loop data analysis only refer to the behaviour at
one cross section of the motorway. Although this approach has the advantage
that a huge amount of data is available on road user behaviour in real traffic,
the dynamics of car-following behaviour cannot be studied this way. To study
car-following behaviour of drivers over time and to have full control over the
experimental conditions, a driving simulator study has been conducted on car-
following behaviour in both good and adverse visibility conditions (day/night,
fog) (Hogema and van der Horst, 1994). Visibility distances were 40,80,120 and
600 m, according to the standard definition for the meteorological visual range
(MVR) (White and Jeffery, 1980). Since at night, the rear lights of a vehicle are
visible over a larger distance than the contour or the road outline following the
definition of MVR, the rear lights of the lead vehicle were made visible over a
larger distance according to the results of Heiss (1976). In each run, subjects were
partly free driving (i.e., no lead cars) and partly in car-following situations with
varying speeds. To prevent overtaking, the left lane of the freeway was closed by
means of diagonally striped work-zone panels.
The free driving speeds as found in the simulator experiment reveal a good
resemblance with real-world data in good and very poor visibility. In moderate
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 247

FIGURE 13.16. TTC gas as compared 35 - - - - - - - - - -.......- - .

to TTC stim as a function of visibility,
= TTCstim
light condition, and relative speed.
30 = TTCgas

o= d~y, -29 km/h!

+

25 + = night,
[] = day, -40 .:,+
)( = night," .
~20
u
~ 15 x

.X
10

OL..-_--L._ _.....a.-_ _..a.-_........_ _""""

o 40 80 1 20 160
visibility (m)

visibility, speeds in the simulator appear to be somewhat lower. Driver's high

expectancy of the presence of a lead vehicle in the simulator and the absence of
overtaking possibilities due to the working zone situation may well explain this
difference. The relevant cue in approaching a lead vehicle is TTC: in the simulator
the lead vehicles would become visible at the pre-determined visibility distance,
whereas the speed difference was either 20 or 40 kmlh. TTC stim represents the
corresponding TTC values. In the night conditions, TTC stim values were larger
since, with an equal speed difference, the visibility distance (of the rear lights) was
larger. Figure 13.16 gives a comparison of TTC stim and TTC gas , being the TTC at
the moment the subject fully releases the gas pedal for the first time after the lead
car has become visible.
The finding that the TTC gas curves in the night condition coincide with the day-
time curves when taking into account the different visibility distances for vehicle
contours and rear lights, indicates that, at night, subjects already react to the lead
vehicle before its contour becomes visible. Their first reaction is based on the
extra visibility range provided by the rear lights. Apparently, TTC gas is mainly
determined by whatever is visible first.
The relationship between TTC gas and TTC stim can be described by a linear
regression line (r == 0.97, p < 0.0001) of the following form (Fig. 13.17).
TTC gas == -1.15 + 0.83 x TTC stim (1)
Similar results were obtained for TTC min , the minimum TTC value as reached
during the whole approach phase (Hogema and van der Horst, 1994). Equation (2)
gives the linear relationship between TTC min and TTC stim (r == 0.92, p < 0.0001).
TTC min == -2.58 + 0.80 X TTC stim (2)
248 van der Horst

FIGURE 13.17. TTC gas as a 40 r-------r----r----r--.,..------,.---,----y----,

function of TTC stim according to

Eq. (1).

30

en
ctS
8' 20
l-
I-

10

10 20 30 40
TTCstim

13.4.1.1 TTC and Collision Avoidance Systems (CAS)

The study conducted by van der Horst (1990) provided values for minimum TTC
values that should be avoided in normal traffic conditions. Together with a 1.5-s
reaction time (including the time needed for moving the foot from the gas to the
brake pedal), an average TTC br of 2.5 s would result in a TTC-criterion of at
least 4-s for activating a collision avoidance system (CAS). The first tests with
various CASs reveal that a CAS with this 4-s TTC-criterion in combination with
a system action to the driver via an active gas pedal reduces the percentage of
small headways considerably without having contra-productive effects on other
behavioural measures (Janssen and Thomas, 1997). Apparently, a criterion based
on TTC seems most in line with what drivers expect to get from a CAS. The question
arises whether it would be worthwhile to try other TTC-criterion values as well.
For determining an appropriate criterion setting for a CAS, it would be of interest
to combine these relationships with earlier findings of critical TTC min values. A
TTCmin value of 1.1 s as found in the study on TTC and driver's decision making
in braking, and the 1.5 s TTC min to distinguish critical from normal behaviour in
Traffic Conflicts studies (van der Horst, 1990). By applying Eq. (2), it would result
in a TTC criterion for activating a CAS of about 4.5 and 5 s, respectively (van der
Horst and Hogema, 1994).

13.4.2 Time-to-Intersection (TTl)

van der Horst (1990) describes a study in which encounters between vehicles
approaching from the minor road and traffic from the left on the main road were
analysed quantitatively on the basis of video recordings at a yield intersection.
Figure 13.18 gives the individual time-to-intersection (TTl) curves (TTl defined
by the distance to the edge of the main road divided by the speed at any given time
 13. Time-Related Measures forModelling Risk in Driver Behaviour 249

~
....
.....
.....

-14 - 12 - 10 -8 -6 -4 -2 o
time (5)

FIGURE 13.18. Individual TTl curves of straight-going cars from theminor road at a yield
intersection involved in an encounter with traffic from theleft.

moment) of straight-goingcars from the minor road. An approach and negotiation

of the intersection with a constant speed would result in a TTl curve linearly
decreasing with time.
By decelerating gradually, it is possible to reduce the decrease of TTl, to keep
TTl constant for a while or even to increase TTL A complete stop would result in
a minimum TTl value (TTlmin) followed by TTl increasing to infinity. Again, it
is striking that hardly any encounter results in a TTlmin < 1.5 s (Fig. 13.18). This
indicates that drivers from the minor road display rather consistent behaviour in
an encounter with another road user and that they may use a time-related measure
such as TTl directly for controlling the braking process relative to the intersection
geometry.

13.4.3 Time-to-Stop-Line (ITS)

A description of drivers' decision making in terms of a time-related measure
such as time-to-stop-line (TTS) has been successfully applied in studies on driver
behaviour at signalised intersections and railway grade crossings. It accounts for
the individual approach speeds of vehicles (van der Horst and Wilmink, 1986;
Tenkink and van der Horst, 1990). Figure 13.19 shows that the willingness to
stop at AKI railway grade crossings (controlled by automatic signal control and
bell sound, no gates) is higher than at signalised intersections. Also the type of
intersection control has its influence on drivers' decision making. At vehicle-
actuated control the occasion where a driver is able to run a red light is much
lower than at fixed-time controlled intersections. However, the willingness to stop
in case of being confronted with the onset of yellow appears to be lower. Drivers
have certain expectations about the functioning of the vehicle-actuated control
system and act accordingly.
250 van der Horst

1.0t----------t~~~ ...-------I
0\
c
'0. 0.8
c-
o
.....
: 0.6
o
>. AKI railway grade
~ 0.4 crossings
J:J signaUsed intersections
.8 vehicle-actuated control
2 0.2 - - - fixed time
0. -log linear model fit

O......----....-L--------------~
o 2 345 6 7
TTS (5)

FIGURE 13.19. Probability of stopping for AKI railway grade crossings (Tenkink and van
der Horst, 1990) versus vehicle-actuated controlled and fixed-time control intersections
(van der Horst and Wilmink, 1986, and Williams, 1977, respectively).

13.5 Conclusions
On the basis of the described studies, it can be concluded that one can learn a lot
from systematic behavioural observations in real traffic to come up with measures
for improving both safety and efficiency. These studies also point to the direct use
of time-related measures such as TTC and TLC as a direct cue for decision making
in longitudinal and lateral control of the vehicle. Time-related measures such as
TTl and TTS serve as appropriate measures for modelling driver behaviour when
negotiating intersections.
These time-related measures provide a framework for modelling driving as a
supervisory control task. Driving is considered to be a time-management task in
which two related measures at the different levels of the driving task are compa-
rable and may serve as a uniform decision criterion when to switch among sub-
tasks.
One example for modelling driver behaviour for automotive applications is an
appropriate criterion for activating a CAS (collision avoidance (warning) system)
based on the TTC approach, together with results of studies on driving behaviour
in adverse visibility conditions resulting in TTC criterion settings in the range
of 4.5-5 s. For driving in fog as an important application area of CASs, special
attention should be given to a distance range of 40-120 m.

References
Alexander, G.J. and Lunenfeld, H. (1986). Driver expectancy in highway design and
traffic operations (Report No. FHWA-TO-86). Federal Highway Administration,
Washington, DC.
 13. Time-Related Measures for Modelling Risk in Driver Behaviour 251

Allen, T.M., Lunenfeld, H. and Alexander, G.J. (1971). Driver information needs. Highway
Research Record, 366, 102-115.
Godthelp, 1. (1984). Studies on human vehicle control. Ph.D. Thesis, Delft University of
Technology, Delft.
Godthelp, J. (1988). The limits of path error neglecting in straight lane driving. Ergonomics,
31,609-619.
Godthelp, 1. and Kaeppler, W.D. (1988). Effects of vehicle handling characteristics on
driving strategy. Human Factors, 30, 219-229.
Godthelp, 1., Milgram. P. and Blaauw, G.J. (1985). The development of a time-related
measure to describe driving strategy. Human Factors, 26, 257-268.
Hayward, 1.C. (1972). Near miss determination through use of a scale of danger (Report
no. TTSC 7115). The Pennsylvania State University, Pennsylvania.
Heiss, W.H. (1976). Highway fog visibility measures and guidance systems (NCHRP Report
171). Transportation Research Board, Washington, DC.
Hogema, 1.H. and Horst, A.R.A. van der (1994). Driving behaviour under adverse visibility
conditions. In ERTICO (Ed.). Towards an intelligent transport system: Proceedings ofthe
First World Congress on Applications of Transport Telematics and Intelligent Vehicle-
Highway Systems (pp. 1623-1630). Artech House, Boston London.
Horst, A.R.A. van der (1990). A time-based analysis of road user behaviour in normal and
critical encounters. Ph.D. Thesis, Delft University of Technology, Delft.
Horst, A.R.A. van der (1991a). Video analysis of road user behaviour at intersections. In
T.W. van der Schaaf, D.A. Lucas and A.R. Hale (Eds.). Near miss reporting as a safety
tool (pp. 93-109). Butterworth-Heinemann Ltd, Oxford.
Horst, A.R.A. van der (1991b). Time-to-collision as a cue for decision-making in braking ..
In A.G. Gale et al. (Eds.). Vision in vehicles III (pp. 19-26). Elsevier, Amsterdam.
Horst, A.R.A. van der (1996). Speed-reducing measures for 80 kmlh roads. In Proceedings
of the 9th ICTCT Workshop in Zagreb (pp. 84-95). University of Lund, Lund, October
1996.
Horst, A.R.A. van der (1998). Factors influencing drivers' speed behaviour and adaptation
(TNO-Report TM-98-D006). TNO Human Factors, Soesterberg.
Horst, R. van der (2004). Occlusion as a measure for visual workload: An overview of TNO
occlusion research in car driving. Applied Ergonomics, 35, 189-196.
Horst, A.R.A. van der and Hoekstra, W. (1994). Testing speed reduction designs for 80
kilometer per hour roads with simulator. In Transportation Research Record 1464 (pp. 63-
68). Transportation Research Board, Washington, DC.
Horst, A.R.A. van der and Hogema, J.H. (1994). Time-to-collision and collision avoidance
systems. In Proceedings 6th ICTCT Workshop (pp. 109-121). Salzburg. Kuratorium fur
Verkehrssicherheit, Salzburg, October 27-29, 1993.
Horst, A.R.A. van der and Wilmink, A. (1986). Drivers' decision-making at signalised
intersections: An optimisation of the yellow timing. Traffic Engineering and Control,
27(12), 615-622.
Hyden, C. (1987). The development of a method for traffic safety evaluation: The Swedish
Traffic Conflicts Technique (Bulletin 70). University of Lund, Lund Institute of Technol-
ogy, Department of Traffic Planning and Engineering, Lund.
Janssen, W.H. and Thomas, H. (1997). In-vehicle collision avoidance support under adverse
visibility conditions. In I. Noy (Ed.). Ergonomics and safety ofintelligent driver interfaces
(pp. 221-229). Lawrence Erlbaum, Mahwah, N1.
Lee, D.N. (1976). A theory of visual control of braking based on information about time-
to-collision. Perception, 5, 437-459.
252 van der Horst

Milgram, P. and Horst, R. van der (1984). Field-sequential colour stereoscopy with liquid
crystal spectacles. In Proceedings Fourth International Display Research Conference.
Societe des electriciens, des electroniciens et des radio-electroniciens, Paris.
Rasmussen,1. (1985). Trends in human reliability analysis. Ergonomics, 28(8),1185-1195.
Tenkink, E. and Horst, A.R.A. van der (1990). Car driver behavior at flashing light railroad
grade crossings. Accident Analysis and Prevention, 22(3), 229-239.
Theeuwes,1. (1993). Visual attention and driving behaviour. In Proceedings ofthe Interna-
tional Seminar Human Factors in Road Traffic. 5-6 April 1993. Universidade do Minho,
Braga, Portugal.
White, M.E. and Jeffery, DJ. (1980). Some aspects ofmotorway traffic behaviour in fog
(TRRL laboratory report 958). Transport and Road Research Laboratory, Crowthorne.
Williams, W.L. (1977). Driver behavior during the yellow interval. In Transportation Re-
search Record 644 (pp. 75-78). Transportation Research Board, Washington, DC.
Winsum, W. van, Brookhuis, K.A. and Waard, D. de (2000). A comparison of different
ways to approximate time-to-line crossing (TLC) during car driving. Accident Analysis
and Prevention, 32,47-56.
14
Situation Awareness and Driving:
A Cognitive Model
M. BAUMANN AND J. F. KREMS

14.1 Introduction
One of the major preconditions of safe driving is that drivers correctly perceive and
interpret the relevant objects and elements of the current traffic situation and that
they consider these elements in planning and controlling their behaviour. Such
elements may be other drivers, the condition of the street or traffic signs. For
each of these elements drivers do not just have to perceive them but they must
understand them according to their relevance to their goals. In addition, drivers
must also make assumptions about the future actions or states of these elements.
For example, perceiving a car coming from the right when entering a crossroads is
far from being enough in order to react accordingly. The driver must interpret this
car according to its relevance to his own goal, that is, safely passing the crossroads.
He has to take into account whether he or the car from the right has to give way.
But even this is not enough to select the appropriate action. If the other car has to
give way, the driver will try to assess from the speed of the car whether the other
car will indeed stop. A concept that has recently become rather popular in aviation
psychology and that aims at describing and integrating these different cognitive
processes is called situation awareness.

14.2 Situation Awareness

According to Endsley (1995b), situation awareness comprises a state of knowl-
edge, that is, knowing what is going on. She uses the term synonymously with
situation model, defining situation awareness as 'the perception of the elements
in the environment within a span of time and space, the comprehension of their
meaning and the projection of their status in the near future' (p. 36). According
to this definition, three levels of situation awareness can be distinguished. Level
I involves perception of the status, attributes and dynamics of the relevant situa-
tion elements. Level II, the comprehension level, involves integrating the different
situation elements into a holistic picture of the situation, resulting in the compre-
hension of the meaning of the different elements. Level III, situation assessment,

253
254 Baumann and Krems

involves the generation of assumptions about the future behaviour of the elements
on the basis of the comprehension of the situation. Situation assessment encom-
passes the processes of achieving, acquiring and maintaining situation awareness
and is distinct from it.
Situation awareness as a state of knowledge and situation assessment as the
processes that lead to this state of knowledge are highly interdependent. Adams
et al. (1995) describe this interdependence, using Neisser's (1976) perception-
action cycle. Perceived changes in the task environment may lead to a change
in the comprehension of the current situation, that is, to a change in situation
awareness. This updated situation awareness will trigger certain actions that might
lead to the perception of new pieces of information, that is, because these actions
change the task environment or because they comprise the sampling of new pieces
of information. In either case the perception of these new pieces of information
will again change the current situation model. This will again trigger new actions
and so on. This is in line with Klein's (2000) proposal that constructing situation
awareness is an active process of guided information seeking rather than a passive
receipt and storage of information. Klein points out that situation awareness is
determined not only by the current situation but also by the person's actions.

14.2.1 An Algorithmic Description of Situation Awareness

Endsley's (1995b) model of situation awareness is certainly one of the most in-
fluential models of situation awareness to date. A number of studies on situation
awareness as well as efforts in modelling focus on different aspects of situation
awareness. Most of them stay within the scope and constraints of Endsley's model
(Rousseau et al., 2005). However, despite the remarkable influence of the model,
it also has some shortcomings. Endsley's model is a descriptive model of situation
awareness; that is, it aims at describing situation awareness in terms of the pro-
cesses that serve it. According to Marr (1982) and Dawson (1998), an information
processing system can be analysed at three levels: Computational, algorithmic and
implementational. The level of concern for descriptive models of situation aware-
ness is the algorithmic level. At this level of analysis an information-processing
system is described in terms of information-processing steps and a functional ar-
chitecture. In the end the goal at this level is to identify the connection between
the functional description of the information processor and its implementational
description. This is achieved when the computational steps an information pro-
cessing system takes can be described in terms of primitive functions that are
implemented in the system's hardware. In this sense the three levels of situation
awareness in Endsley's model can be viewed as different functions of situation
awareness (McGuinnes and Foy, 2000, cited in Rousseau et al., 2005). Yet even
when Endsley (1995b) describes some more primitive mechanisms of how these
functions are performed, these mechanisms are far too abstract to be part of the
functional architecture. The goal of this paper is to move the algorithmic analy-
sis and description of situation awareness a step further, to a description in terms
of the functional architecture and therefore to come closer to a more complete
 14. Situation Awareness and Driving: A Cognitive Model 255

algorithmic analysis of situation awareness. To this end, we propose a model that

describes how situation awareness is (1) constructed, (2) maintained and used as
a basis for (3) action selection in terms of primitive functions.
The present model views the construction and maintenance of situation aware-
ness as a comprehension process. The model proposes in detail a mechanism of
how the comprehension of the situation, that is, how Level II situation awareness is
achieved. Essentially it assumes that the comprehension process for understanding
the meaning of different elements in a situation is comparable to the comprehension
process for understanding language or discourse. In both cases an integrated mental
representation of the perceived and processed pieces of information is constructed.
This representation reflects the understanding of these elements. Fundamental to
our approach is Kintsch's (1998) Construction-integration theory of text compre-
hension. As was explained above, situation awareness is a dynamic representation
that is influenced by both the situation and the actions of the person. Therefore,
for a model of situation awareness to be complete it has to specify how situation
awareness is translated into action. As for the construction of situation awareness
we believe that cognitive psychology already offers a theory that can be used for
this purpose. The theory we use here is Norman and Shallice's (1986) theory of
action selection. Groeger (2000) described how this theory can be applied within
the driving context to explain the selection and execution of driving manoeuvres.
In this paper we will describe how this theory can be combined with Kintsch's
construction-integration theory to provide a comprehension based model of sit-
uation awareness. With the combination of these two theories it is possible to
describe in more detail than in Endsley's model how situation awareness is con-
structed, maintained and used as a basis for action selection.

14.2.1.1 The Construction of the Situation Model: Comprehending the Situation

How is the situation model built according to the construction-integration the-
ory? According to Kintsch (1998), comprehending new pieces of information in-
volves two phases. In the first phase, the construction phase, the perception of new
pieces of information activates knowledge structures in long-term memory, such
as propositions and schemata that are associated with these pieces of information.
This activation process is undirected and follows learned associations among these
different pieces of knowledge. The result of this automatic activation process is a
rather unstructured activation of knowledge in long-term memory. This activated
knowledge then becomes integrated in a following step, the integration phase. In-
tegration is accomplished through inhibitory and excitatory processes: Knowledge
that has become activated in the first phase and is compatible to already activated
knowledge (i.e., to the existing situation model) will remain activated and at the
same time activate other knowledge compatible with it. Simultaneously, incom-
patible pieces of knowledge inhibit each other. For example, if an event such as
a traffic light turning yellow triggers two incompatible interpretations, 'I have to
decelerate to stop before the traffic light' and 'I have to accelerate to pass the
crossroads before the traffic light turns red', these two interpretations will inhibit
256 Baumann and Krems

each other. At the same time, these two interpretations are also activated or inhib-
ited by other knowledge. For example, if the driver knows that the police monitor
this crossroads, that knowledge will inhibit the acceleration interpretation. How-
ever, being a short distance from the traffic light will activate the acceleration
interpretation. In the end, the 'winning' interpretation will be the one that receives
the most activation from other activated knowledge.
The result of these two phases is an episodic memory representation. Despite
being a unitary representation, Kintsch (1998) distinguishes between two com-
ponents of the episodic memory representation: The text base and the situation
model. The text base consists of information that is directly derived from the
text. The situation model results from the connection between the text represen-
tation and the comprehender's world knowledge. Translating into the domain of
driving and situation awareness, we propose that the episodic memory represen-
tation similarly consists of two components: A situation-specific representation
(analogous to the text base model) and the situation model. The situation-specific
representation consists of information that is directly perceived from the envi-
ronment. This mainly includes the situation elements the driver perceives and
information about the status of these elements that can be directly perceived,
such as one's own speed. It also might include inferences that can be directly
made from the perception of the situation elements and their status, such as that
one's own speed is greater than the speed of the lead vehicle. This representa-
tion might reflect what Endsley (1995b) terms Level I situation awareness. The
situation model then results from the connection between this situation-specific
representation and the driver's world knowledge. Through this connection the per-
ceived configuration of situation elements becomes connected to prior experienced
situations with similar configurations of elements. This connection provides an in-
terpretation of the current situation, partly depending on the outcome of similar
prior situations. This mechanism might at least in part explain the advantage of
experienced drivers compared to novice drivers in hazard perception (Chapman
and Underwood, 1998; Groeger, 2000; McKenna and Crick, 1994). Experienced
drivers have a much greater chance of having stored in long-term memory the rel-
evant information that identifies the current situation as dangerous because their
database of experienced traffic situations is much greater than that of novices. But
the situation model does not only provide interpretations of the current situation,
such as whether it is dangerous or not. It also contains expectations about the
future development of the situation and the future behaviour of the situation ele-
ments. These expectations are also derived from prior experienced situations that
get connected to the situation-specific representation. For example, an experienced
driver will know how long it will take a yellow light to tum red and perceiving
a traffic light turning yellow will automatically activate this expectation. The sit-
uation model therefore might reflect Endsley's Level II and Level III situation
awareness.
To summarise, this model proposes that the different levels of situation aware-
ness are highly interconnected as they are part of an unitary episodic memory
representation. By using Kintsch's (1998) construction-integration theory of text
 14. Situation Awareness and Driving: A Cognitive Model 257

comprehension, the model also specifies a mechanism of how this representation

is constructed beginning with the perception of elements of the current situation.
It is accomplished by a two-phase process. The first phase consists of an unstruc-
tured activation process, the second phase of a constraint-satisfaction process. This
constraint-satisfaction process keeps the representation of the current situation co-
herent through the reciprocal activation of compatible knowledge and inhibition
of incompatible knowledge.

14.2.1.2 Selection of Actions and the Control of Behaviour

The selection of situation-appropriate actions can also be integrated seamlessly
into this comprehension-based model of situation awareness. During the construc-
tion of the situation representation, not only knowledge relevant to the interpre-
tation of the situation and its future development get activated, but also actions
that are appropriate in the current situation. According to the framework of Nor-
man (1981) and Norman and Shallice (1986), actions are represented as schemata.
These schemata are programmes for the control of routine tasks. Each schema is
part of a hierarchy of schemata and is connected to superordinate and subordinate
schemata. For example, the schema 'stop at a red traffic light' is subordinate to a
general driving schema and it is superordinate to other schemata, such as 'push the
brake pedal to decelerate'. Besides these hierarchical connections there are also
connections to functionally related schemata or to schemata that are incompatible.
The connections are either excitatory, like the connections from a superordinate
schema to its subordinates, or inhibitory if the schemata are incompatible with
each other, such as accelerating or decelerating when facing a yellow traffic light.
Each schema is connected with certain trigger conditions that activate the schema
if present in the current situation representation. In our example, detecting a yellow
traffic light activates both the acceleration and deceleration schemata. In this way,
the activation of situation-appropriate actions becomes part of the construction
of the situation representation. One possibility of how these trigger conditions
become connected to action schemata is through prior experience. Successfully
performing an action in a given situation will establish excitatory links between
relevant elements of the situation and this action schema. These elements will then
become the trigger conditions for this schema. A failure in performing this action
in this situation may reduce the strength of already existing excitatory connections
or might even establish inhibitory links between relevant situation elements and
the action schema. If more than one schema is activated, as in the above example,
a selection has to be made. Norman and Shallice (1986) propose a mechanism
called contention scheduling, which operates on two processes: Competition, the
different action schemata compete with one another for their activation value, and
selection, which takes place on the basis of the activation alone. In the traffic light
example, the yellow traffic light triggers two possible action sequences, decelerat-
ing and accelerating. Both schemata receive activation as their trigger conditions
are satisfied. At the same time, other elements in the situation may be present
such as a leading car that is already decelerating. This additional element will
258 Baumann and Krems

activate the deceleration schema further. Because of the inhibitory link between
the deceleration and acceleration schema, both schemata inhibit each other. As the
deceleration schema becomes more activated than the acceleration schema, the
acceleration schema will be more strongly inhibited. As a result, the deceleration
schema is more highly activated than the acceleration schema and the deceleration
schema takes over the control of action. In sum, contention scheduling is sufficient
to control well-learned, simple action sequences. It resolves competition for se-
lection, prevents competitive use of common resources and negotiates cooperative
use of common resources when possible.
In cases where a schema is not available or when the task is novel or com-
plex, another control structure is necessary. Norman and Shallice (1986) call this
structure the Supervisory AttentionalSystem(SAS). The SAS allows for voluntary,
attentional control of performance. The SAS is basically an attentional system
that influences the selection of schemata in the contention scheduling process by
providing additional activation and/or inhibition to schemata in order to bias their
selection. The selection process itself is always based on the highest activation
values of the competing schemata. This system allows for top-down influences in
the action selection process. For example, if a physician on the way to an emer-
gency call sees the traffic light turning yellow, he will not stop. Instead he will try
to proceed even if the light is red because of his goal to get to the patient as quickly
as possible. In this case the deceleration schema receives greater inhibition from
the output of the SAS and is unlikely to be selected by the contention scheduling
process.
To summarise, we described two well-established theories, Kintsch's (1998)
construction-integration theory of text comprehension and Norman and Shallice's
(1986) theory of action selection. Both theories possess a firm empirical foundation
and describe their assumptions in detail. In addition, Norman and Shallice's theory
was already applied to the driving context as a framework to describe the selection
and execution of driving manoeuvres (Groeger, 2000). What we believe is new
and fruitful is the combination of these two theories as a framework for situation
awareness. We proposed to combine these two theories into a comprehension-
based model of situation awareness. By this it is possible to make much more
detailed assumptions about the processes that are involved in the construction and
maintenance of situation awareness and its use as a basis for action selection in
driving.

14.3 Errors and the Comprehension Based-Model

of Situation Awareness
Driver distraction is a major cause for errors in driving. Distraction can stem from
different sources, such as other occupants, objects outside the car or in-vehicle
tasks. The distraction can be visual, as when the driver looks at an in-vehicle display
while driving; psychomotor, for example when the driver removes a hand from
the wheel to adjust the heat; auditory, such as when the radio obfuscates auditory
 14. Situation Awareness and Driving: A Cognitive Model 259

signals from other vehicles; or cognitive, for example, when the driver's mind is
preoccupied with a conversation. The comprehension-based model of situation
awareness addresses especially the last kind of distraction, cognitive distraction.
Next we will demonstrate how certain errors in driving can be explained within
the framework of this model.
An error that is called 'inattention blindness' or 'looked-but-did-not-see' de-
scribes the phenomenon that the distracted driver was actually looking at the rele-
vant situation element (e.g., the red traffic light), but did not react accordingly. How
can this be explained in terms of our model? The comprehension-based model of
situation awareness implicitly assigns working memory a key role in the processes
of situation awareness construction, maintenance, updating and action selection.
Working memory resources are necessary for associating perceived elements in
the environment with knowledge stored in long-term memory, for integrating these
new elements in the current situation model, for removing irrelevant elements from
the situation model, for keeping the information in the situation model available
for the selection of appropriate actions, for monitoring the selection and execution
of actions and so on. In the case of cognitive distraction, some of these resources
may not be available because they are assigned to other tasks, such as entering the
destination into a navigation system or talking to the passenger. The remaining
working memory capacity might be too reduced to ensure that the perceived situa-
tion elements get fully connected to the relevant knowledge in long-term memory.
The comprehension process stops prematurely and some relevant implications of
the perceived situation element do not become activated and therefore are not
available in the situation representation to trigger the relevant action schemata. In
addition, the impoverished processing of a perceived situation element may lead
to a shallow memory trace for this element, which then rapidly decays and is lost
from the situation representation. Again, this situation element is not considered
in the action selection process.
Cognitive distraction may not only impair the comprehension of situation ele-
ments, but it can also impair action selection directly. As explained above, actions
are controlled by action schemata that compete for the highest activation value.
The activation of each schema is partly determined by the match between its trigger
conditions and the current situation and by the SASe The SAS preactivates those
schemata that are compatible with the current task goal and inhibits those that are
incompatible. Yet as an attentional system the SAS depends on the availability of
the respective cognitive resources. If these are not available because the driver is
preoccupied with a demanding in-vehicle task or engaged in a conversation, action
selection may be primarily determined by the match between the schemata's trig-
ger conditions and the current situation. In such a case the driver often exhibits the
behaviour that is most strongly connected to the current situation even if it is not
compatible with the current task goal. As an example, imagine someone driving
along a very familiar road. At a certain crossroads the driver is used to turning
right, in the direction of the workplace. But on this day the driver wants to go to the
supermarket, requiring a left tum. The driver is very busy with thinking about what
to buy for dinner and automatically turns right, towards the workplace, instead of
260 Baumann and Krems

to the supermarket. In this case the resources of the SAS were so occupied with
reasoning about what to buy that it could not no longer control behaviour. That
is, preactivation of the less common action schema for this situation ('tum left')
failed in order to prevent the most highly connected action schema to this situation
('tum right') from taking over control.

14.4 Situation Awareness and In-Vehicle Information System

Tasks

One of the major concerns connected to the on-board use of information and
driver assistance systems is related to this kind of driver distraction and the errors
that follow from impaired situation awareness. Safe driving requires uninterrupted
surveillance of one's vehicle and awareness of the traffic situation and changes in
environmental conditions. Interacting with an in-vehicle information system can
lead to distraction, increased cognitive workload and a rise in the probability of
accidents. Therefore, within the ITS domain there is a need for a procedure to
measure the effects of IVIS tasks on the driver's situation awareness (European
Commision, 2000).
The theory outlined above emphasises the importance of working memory pro-
cesses for the construction and maintenance of the situation representation while
driving. It immediately follows that IVIS tasks that are associated with high work-
ing memory load should clearly impair the construction and maintenance of a
proper situation representation.

14.4.1 A Measurement Procedure: Context-Dependent

Choice Reaction Task
How can the effect of IVIS tasks on situation awareness be measured? One pos-
sible way is to have drivers perform these IVIS tasks while driving and use some
measure to determine their effect. However, up to now most of the available pro-
cedures for the measurement of situation awareness were developed for use in the
aviation domain. Situation Awareness Global Assessment Technique (SAGAT;
Endsley, 1995a) was developed for investigating pilots' situation awareness. HauB
and Eyferth (2001) developed SALSA, a similar technique used in air-traffic con-
trol studies. Only recently have procedures for studying the driving task begun to
be developed. For example, Bailly et al. (2003) propose a methodology that uses
video to analyze drivers' situation awareness.
Both SAGAT and SALSA are so-called 'freezing techniques'. Domain experts,
such as pilots, perform the relevant task in a simulator. While working on the task
(e.g. a flight mission), the simulation is halted and the experts are asked several
questions regarding relevant parameters of the current task situation. The propor-
tion of questions answered correctly is used as an indicator of the pilot's situation
awareness. The advantage of these procedures is that they directly measure the
 14. Situation Awareness and Driving: A Cognitive Model 261

availability of relevant parameters of the current task. However, these techniques

require, as a prerequisite, a detailed task analysis that can be used to formulate the
questions asked during the task interruption. These techniques are content based
in so far as they compare the content of the operator's situation awareness with the
obligatory content derived from the task analysis. Given the diversity of driving
tasks and driving situations as well as the dynamics of the driving task, such a
task analysis is very difficult to perform within the driving domain. Furthermore,
it is difficult to draw conclusions about the causes of degraded situation awareness
using these procedures.
The focus of the measurement approach we propose is on the processes that
serve to maintain situation awareness in terms of the comprehension-based model
of situation awareness. Therefore this approach could be characterised as a process-
based approach of situation awareness measurement. The basic idea is to measure
the degree of interference between the processes involved in the construction and
maintenance of situation awareness and those that are involved in performing IVIS
tasks. In terms of our model of situation awareness, both visual attention and work-
ing memory playa key role for situation awareness. Therefore, our procedure uses a
dual-task technique to measure the visual attention and working memory demands
of different IVIS tasks. IVIS tasks are performed in the laboratory and regarded as
the primary task. A secondary task is used to measure the residual visual attention
and working memory resources not utilised by the primary IVIS task. As men-
tioned above, the underlying assumption is that IVIS tasks that demonstrate high
demands for visual attention and working memory interfere with the construction
and maintenance of a proper situation representation in memory and therefore will
lead to degraded situation awareness.
This secondary task is a choice reaction task where the appropriate reactions to
the different stimuli are context-dependent. It combines a visual perception task
with a memory task. The perceptual component of the task consists of presenting
one of two possible visual stimuli at 23 0 of visual angle, either to the right or
to the left of the participant at a distance of 2.25 m. The stimuli are presented
with interstimulus intervals varying between 1 and 3 s. The participant's task is to
respond with a button press. This component of the task is similar to a laboratory
variant of the peripheral detection task (PDT). Winsum et al. (1999) developed the
PDT on the basis of Miura (1986) and Williams (1985,1995). Miura (1986) found
that reaction times to spots of light presented at different horizontal eccentricities
on the windscreen while driving increased with traffic density, thereby reflecting
demands of the driving task. Williams (1985, 1995) showed that with increasing
foveal load the accuracy of responses to stimuli presented peripherally decreased.
Within the advanced driver attention metrics (ADAM) project (Baumann et al.,
2003; Breuer et aI., 2003), we developed a version of the PDT that is applicable
under laboratory conditions. It has been shown that this version is a valid measure
of the visual demand of IVIS tasks (Baumann et aI., 2003).
But contrary to the PDT the procedure used in this project to measure the
visual and the cognitive demand of IVIS task requires a choice reaction not only
a detection reaction. The participants have to press a different button depending
262 Baumann and Krems

on the stimulus presented. But the correct response depends not only from the
presented stimulus but also from the current context. For example in context A,
the participant presses the right button if stimulus 1 appears and the left button
if stimulus 2 appears. In context B the stimulus-button pairing is reversed. The
current context is signalled by acoustic signals, a high- and a low-frequency tone.
Every three to five visual stimuli a new context signal is presented that can indicate
either a new context or the same as before. Remembering the current context and
updating the context define the memory part of this task.
This task yields two measures that are intended to allow the independent assess-
ment of visual and working memory demands of the primary task. First, one can
look at the proportion of visual stimuli responded to, not differentiating whether
the response was correct or not according to the respective context. This detection
rate is used as a measure of the visual demand of the primary task. If the primary
task is highly visually demanding, the participant will simply miss more of the
visual stimuli than if the primary task is of low visual demand. The proportion of
correct reactions from all given reactions is used as a measure of working memory
demands of the primary task. If the task is highly demanding in terms of working
memory, the participants may often forget the current context or may frequently
fail to update the context after a new context signal was presented. This leads to
more erroneous responses to the visual stimuli.

14.4.2 Evaluation ofthe Context-Dependent Choice Reaction

Task
In a current project the validity of the context -dependent choice reaction task is
evaluated. The first step consists of evaluating differently visually and cognitively
demanding tasks with this procedure. The aim is to examine whether the detection
rate and the correct response rate are sensitive to visual and working memory de-
mands of IVIS tasks. Specifically, we wanted to know whether the simple detection
rate is sensitive primarily to the visual demand of the IVIS task and whether the
correct response rate is sensitive primarily to the memory demands of the IVIS
task. The second step of the project consists of comparing these evaluations of
visual and working memory demands of IVIS tasks with the effects of these IVIS
tasks on situation awareness while driving. In this paper we will summarise results
of the first experiment.
In this experiment, 23 participants had to perform either a demanding visual
search task or a demanding working memory task as primary tasks. The visual
search task consisted of finding a target letter among 160 distractor letters. The
working memory task consisted of a one-back task. In this task participants were
shown a series of letters, one at a time. Each time a new letter was presented
participants had to name the letter that was presented just before. Simultane-
ously to these tasks, participants performed the context-dependent choice reaction
task. As a baseline condition participants performed the context -dependent choice
reaction task also alone. We expected that the simple detection rate of the sec-
ondary task would be lower when the visual search task was the primary task,
 14. Situation Awareness and Driving: A Cognitive Model 263

TABLE 14.1. Detection and correct response rate in the

baseline, visual and working memory condition.
Detection rate Correct response rate

Mean SD Mean SD

Baseline 0.98 0.05 0.87 0.20

Visual 0.81 0.13 0.81 0.13
Working memory 0.89 0.09 0.78 0.11

than when the working memory task was the primary task. The reverse pattern
was expected for the correct response rate. Table 14.1 presents the mean detection
and correct response rates in the baseline, visual and working memory condi-
tion. The rates are based on the responses to the first stimulus after a context
change.
As predicted, the results show that the detection rate is clearly lower when
the visual search task is the primary task than for the working memory task,
Z == - 2.516, p == 0.012. This indicates that the detection rate is sensitive to the
visual demand of the primary task, confirming the results of Baumann et al. (2003).
The correct response rate was only slightly lower for the working memory task
than with the visual search task, indicating that the correct response rate is not very
sensitive to the working memory demands of the primary task. The difference was
not significant, Z == -0.608, p == 0.543. We assume that the failure to find a
significant difference in the correct response rate between the visual and working
memory condition is due to the too small working memory demands of the context-
dependent choice reaction task. Therefore, in a second experiment we will increase
the working memory demand of the task be increasing the frequency of the context
changes.

14.5 Conclusions
The goal of our approach is to establish the cognitive basis of situation awareness
in order to be able to apply it to the driving task. We assume, similar to Adams et al.
(1995), that the construction of situation awareness is basically a comprehension
process that yields a mental representation of the meaning of the different elements
of a traffic situation and the situation as whole, that is, the situation representation.
This situation representation serves as a basis for planning future behaviour that in
tum alters the situation representation again. According to this perspective, driver
knowledge plays a key role in determining the significance of events and elements
in a traffic situation. Viewing situation awareness as a comprehension process
also highlights the role of working memory in the process of constructing and
maintaining situation awareness. The processes necessary to interpret new pieces
of information, to determine their consequences for the current situation model,
to integrate new pieces of information into the situation model and to remove
264 Baumann and Krems

irrelevant information from the situation model must take place in working mem-
ory. If the resources of working memory are occupied by other tasks, for example
IVIS tasks, these processes will be impaired leading to a degraded situation aware-
ness. This process model of situation awareness is used to develop a procedure
that should allow the evaluation of visual and working memory demands of IVIS
tasks. The idea is that knowing the visual and working memory demands of IVIS
tasks allows to make predictions about the effects of these IVIS tasks on situation
awareness.

References
Adams, MJ., Tenney, YJ. and Pew, R.W. (1995). Situation awareness and the cognitive
management of complex systems. Human Factors, 37(1), 85-104.
Bailly, B., Bellet, T. and Goupil, C. (2003). Driver's mental representations: Experimental
study and training perspectives. In L. Dom (Ed.). Driver behavior and training. Aldershot,
UK, Ashgate.
Baumann, M., Rosler, D., Jahn, G. and Krems, J.F. (2003). Assessing driver distraction using
occlusion method and peripheral detection task. In H. Strasser, K. Kluth, H. Rausch and
H. Bubb (Eds.) Quality of work and products in enterprises of the future. Ergonomia
Verlag, Stuttgart, pp. 53-56.
Breuer, 1., Bengler, K., Heinrich, C. and Reichelt, W. (2003). Development of advanced
driver attention metrics (ADAM). In H. Strasser, K. Kluth, H. Rausch and H. Bubb (Eds.).
Quality of work and products in enterprises of the future. Ergonomia Verlag, Stuttgart,
pp.37-39.
Chapman, P.R. and Underwood, G. (1998). Visual search of driving situations: Danger and
experience. Perception, 27(8), 951-964.
Dawson, M.R.W. (1998). Understanding cognitive science. Blackwell, Malden.
Endsley, M.R. (1995a). Measurement of situation awareness in dynamic systems. Human
Factors, 37(1), 65-84.
Endsley, M.R. (1995b). Towards a theory of situation awareness in dynamic systems. Human
Factors, 37(1), 32-64.
European Commission (2000). Recommendation of 21st December 1999 on safe and ef-
ficient in-vehicle information and communication systems: A European statement of
principles on human machine interface. Official Journal, 2000/53/EC.
Groeger, J. (2000). Understanding driving. Psychology Press, Hove.
HauB, Y. and Eyferth, K. (2001). The evaluation of a multi-sector-planner concept:
SALSA - A new approach to measure situation awareness in ATC. Fourth Interna-
tionalAir TrafficManagement R&D Seminar ATM-2001. (http://atm2001.eurocontrol.fr'/
finalpapers/pap 136.pdf).
Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge University
Press, New York.
Klein, G. (2000). Cognitive task analysis of teams. In 1.M. Schraagen, S.F. Chipman and
T.D. Shalin (Eds.). Cognitive task analysis. Lawrence Erlbaum Associates, Mahwah,
pp. 417-430.
Marr, D. (1982). Vision. Freeman, San Francisco.
McKenna, F.P. and Crick, J.L. (1994). Hazard perception in drivers: A methodology for
testing and training. Transport Research Laboratory (Contractor Report 313), pp. 1-29.
 14. Situation Awareness and Driving: A Cognitive Model 265

Miura, T. (1986). Coping with situational demands: A study of eye movements and periph-
erial vision performance. In A.G. Gale, J.D. Brown, C.M. Haselgrave, P. Smith and S.H.
Taylor (Eds.). Vision in vehicles-II. Elsevier, Amsterdam.
Neisser, U. (1976). Cognition and reality: Principles and implications ofcognitive psychol-
ogy. Freeman, San Francisco.
Norman, D.A. (1981). Categorization of action slips. Psychological Review, 88(1), 1-15.
Norman, D.A. and Shallice, T. (1986). Attention to action: Willed and automatic control
of behavior. In RJ. Davidson, G.E. Schwartz and D. Shapiro (Eds.). Consciousness and
self-regulation (Vol. 4). Plenum Press, New York, pp. 1-18.
Rousseau, R., Tremblay, S. and Breton, R. (2005). Defining and modeling situation aware-
ness: A critical review. In S. Banbury and S. Tremblay (Eds.). A cognitive approach to
situation awareness: Theory and application. Ashgate, Hampshire Burlington, pp. 3-21.
Williams, LJ. (1985). Tunnel vision induced by a foveal load manipulation. Human Factors,
27,221-227.
Williams, LJ. (1995). Peripheral target recognition and visual field narrowing in aviators
and nonaviators. International Journal ofAviation Psychology, 5(2), 215-232.
Winsum, W. van, Martens, M.H. and Herland, L. (1999). The effects ofspeech versus tactile
driver support messages on workload, driver behaviour and user acceptance (Report
No. TM-99-C043). TNO Human Factors. Soesterberg, the Netherlands.
15
Driver Error and Crashes
DIANNE PARKER

15.1 Slips, Lapses and Mistakes

The first distinction to be considered is between slips/lapses and mistakes. Rea-
son (1990) defined error as 'the failure of planned actions to achieve their desired
ends - without the intervention of some unforeseeable event' . In these terms, while
a slip represents a problem with the execution of a good plan, a mistake involves
an inappropriate or incorrect plan that is correctly executed. Slips and mistakes
map directly onto Rasmussen's (1974, 1990) differentiation of three levels of hu-
man performance, According to Rasmussen's model, the cognitive mode in which
people operate changes as the task performed becomes more familiar, from the
knowledge-based through the rule-based to the skill-based level. The three levels
are not mutually exclusive, but represent a progression, leading to skilled per-
fonnance. Knowledge-based performance, which involves consciously thinking
the task through, is relevant when the task faced is novel and conscious effort
must be made to construct a plan of action from stored knowledge. Knowledge-
based performance is necessary if you are planning to drive to a destination never
previously visited. Errors at this level of performance are mistakes, arising from
incorrect knowledge or from the limitations of cognitive resources. Moreover, de-
cision making itself is subject to a range of biases (Parker and Lawton, 2003). For
example, Kahneman and Tverskys' classic laboratory experiments identified the
availability bias, referring to the fact that probability judgements (e.g., judgements
of the likelihood of having a car accident) are strongly influenced by the ease with
which past cases can be recalled (Tversky and Kahneman, 1974).
Rule-based performance occurs when we already have some familiarity with
the task, and can perform it by drawing on a set of stored mental if-then rules. For
example, a learner driver may have learned the appropriate routine for negotiat-
ing a roundabout, but will still probably have to give the task full concentration,
retrieving the rules and applying them appropriately. Errors at this level are also
mistakes, involving the misapplication of a rule, for example misremembering the
correct lane to be taken when turning right at a roundabout. At the skill-based level
of performance tasks are familiar and actions occur without conscious thought.
For example, experienced drivers give little thought to changing gear, steering or

266
 15. Driver Error and Crashes 267

using the brakes and are able to combine these tasks with others such as talking
to a passenger or monitoring hazards on the road ahead. At this level of perfor-
mance, errors come in the form of slips and lapses. For example, driving towards
work when you intend to go to the supermarket on a Saturday because you were
distracted by the children at the point on the route where a different turning was
required.

15.2 Errors and Violations

Having established the distinction between slips/lapses and mistakes, a second
distinction has been made between these types of error and violations. The uni-
fying characteristic of slips, lapses and mistakes is that they are all unintentional.
These different types of error all arise from information-processing problems,
their occurrence can be understood in relation to the cognitive processes of a sin-
gle individual, and their frequency can be reduced by skills training, improving
knowledge, workplace information and redesign.
In contrast, violations often have nothing to do with knowledge or skills and
much more to do with motivation. The chief characteristic of a violation is that it
represents a deliberate deviation from normal or recommended practice: it is, at
least in part, intentional. The frequency of violations cannot usually be reduced by
competence assessment or training, because violations reflect what the individual
decides to do, rather than what he or she can do. A reduction in violations can be
best achieved with attention to aspects of the organisation such as morale, attitudes,
beliefs, norms and organisational safety culture (Reason et al., 1989).
Violations represent deviations from formal or informal rules that supposedly
describe the best/safest way of performing a task, for example a railwayman who
steps on and off a moving shunting engine in order to save time. In this case,
although the action was intended, the occasionally bad consequences were not. For
the most part, violations go unpunished but occasionally and often in combination
with an error, the results can be catastrophic. Violations are of particular interest in
an organisational context where rules, guidelines, policies and protocols are often
developed to control practice (Johnson and Gill, 1993) and prevent mistakes.
Violations can also be sub-divided into types (Lawton, 1998; Reason et al.,
1998). Routine violations occur when an individual believes themselves to have
sufficient skill and experience to violate rules habitually, for example, where fail-
ure to wear a safety helmet is widespread throughout a workforce. Situational
violations occur when the local situation makes following the rules difficult or
impossible. Continuing with the safety helmet example, if an insufficient number
of safety helmets were provided by management, then failing to wear one would
be a situational violation. Optimising violations serve the needs of the violator to
express mastery and skill and may be the province of the very experienced staff
member. Some individuals might choose not to wear a safety helmet perceiving
them to be uncomfortable and embarrassing to wear. Finally, the exceptional viola-
tion occurs when a novel situation arises which cannot be dealt with in accordance
268 Parker

to existing rules. For example, a safety helmet may restrict movement and so
become a safety hazard in a confined area.

15.3 The Manchester Driver Behaviour Questionnaire

Our programme of work has identified three basic types of bad driving, initially
using a 50-item self-report questionnaire, the Manchester Driver Behaviour Ques-
tionnaire (DBQ) (Reason et aI., 1990). The DBQ was administered to a national
sample of 520 drivers, who were asked how often each of the behaviours happened
to them in the course of their normal driving, using a simple frequency response
scale, where 0 == never and 5 == nearly all the time. Factor analysis of their re-
sponses revealed three strong underlying dimensions, reflecting distinct types of
bad driving that were statistically as well as conceptually, different. The question-
naire was streamlined to 24 items, to include the eight top loading errors, lapses
and violations. In a further study, the revised measure was completed by almost
1600 drivers, and the three factors were confirmed.
The first type, true errors, are mistakes and slips, such as misjudging the speed
of oncoming traffic when attempting to overtake, attempting to overtake a vehicle
you had not noticed was signalling a right turn, and failing to notice a Give Way
sign at a junction. This type of behaviour is potentially dangerous, for the driver
and other road users. The second group of behaviours, which we called lapses,
are usually harmless but irritating. They include turning on the windscreen wipers
when you mean to turn on the indicator, forgetting where you have left your car in
a car park, and realising you have no clear recollection of the road you have just
driven along. The focus of this paper is on the third group of behaviours, which we
called violations, includes speeding, tail gating and overtaking on the inside. These
behaviours differ from mistakes, slips and lapses in that they are committed, at
least in part, intentionally and in the knowledge that one is engaging in potentially
dangerous and often illegal behaviour.
Incidentally, this tripartite typology, of violation, error, and lapse has been repli-
cated in studies carried out by other researchers in Britain, and in the Netherlands,
Finland, Sweden, Brazil, Spain, Germany, Australia and China. In particular, Lars
Aberg in Sweden and Timo Lajunen in Turkey have worked with the DBQ ex-
tensively (e.g., Lajunen et al., 2004). While the factor structure uncovered varies
slightly from time to time, there is now clear evidence that this is a robust and
meaningful instrument.

15.4 The DBQ and Road Traffic Accidents

The distinction between error and violation on the road is important because they
are related differently to crash involvement (Parker et al., 1995a). For each of our
sample of 1600 we had access to their driving histories during a specified 3-year
period. We regressed crash rate onto several predictor variables, including the three
 15. Driver Error and Crashes 269

HIGH ACCIDENT RESPONDENTS:

ERROR AND VIOLATION FACTOR SCORES
5
iii
iii
4
m
IiJ
3 II I:J
EJ m m Ii
en raP m m
Z 2
0 E1:I

i= m iii
EEl I!J em
II(
..J
0 lEI
>
0
E1
m Ii
1 EI ~ r:J E:J
EI EI &I

2
-3 -2 -1 0 2 3
ERRORS

FIGURE 15.1. The error and violation factor scores of 185 crash-involved drivers (from
Reason et al., 1989).

DBQ factor scores, and demographic information about the drivers, including their
annual mileage, and age and sex, as these are all known to be associated with crash
involvement.
Annual mileage driven, age and sex were all predictive of crash involvement,
in predictable ways. Those who drove more had more crashes, older drivers had
fewer crashes, and males had more crashes than females. However, even after
statistically controlling for the effects of mileage, age and sex, adding DBQ factor
scores to the equation resulted in a significant improvement in explained variance.
The violations factor score was significantly predictive of crash rates. Equally
significantly, scores on the error factor were not predictive of crash rate. Moreover,
this predictive relationship was found in a subsequent prospective study (Parker et
aI., 1995b) to hold good for crash involvement in the 3-year period after the DBQ
measure was used, showing that the measure is truly predictive.
The relationship is shown in Fig. 15.1, which plots almost 200 high crash-
involvement survey respondents (i.e., those who had had two or more crashes in
the previous 3 years) in terms of their scores on the DBQ violation and error
factors. Factor scores, by their nature, have a mean value of a and a standard
deviation of 1. Therefore, a negative score indicates a higher than average number
of reported errors or violations, and a negative score indicates a lower than average
number. It is evident from Fig. 15.1 that these respondents were characterised by
high mean scores on the violation factor, but not by high mean scores on the
error factor. In the equivalent plot for those respondents who had had only one
270 Parker

crash, error and violation scores are equally distributed around the means on both
factors.
The practical importance of this distinction between errors, on the one hand,
and violations, on the other, is that they have very different psychological origins
and they therefore have very different remedial implications. Errors are based on
perceptual, attentional, or judgmental processes, so if crashes are caused primarily
by errors, we should try to improve the situation by training people to use their
cognitive resources more carefully or efficiently. So errors can be tackled by the
type of driver training most often offered. Violations, on the other hand, would
seem to be based on attitudinal and/or motivational factors. So, if there were
evidence that crashes are caused primarily by violations, it would make sense to
try to improve the situation by changing people's beliefs and/or motives.
It is well established that young males are the highest violators. However, that
piece of information is not all that helpful to road safety professionals. There
will always be young male drivers on the roads, unless some fairly radical policy
decisions are taken. Moreover, no one could seriously argue that there is something
about being young or about being male that directly causes risk-taking behaviour
behind the wheel. It is far more plausible to suggest that there are psychological
factors, which are correlated with youth and maleness, that predispose some drivers
to commit violations. This suggestion is supported by the fact that not all young
males are high violators, and not all high violators are young males.
The crucial characteristic of high violators is that they are choosing to take risks.
The commission of a driving violation is a matter of choice. We do not simply find
ourselves overtaking on the inside, or gesturing to another road user, or shooting
through traffic lights as they tum red. While many drivers stopped by the police
plead that they were unaware that they were breaking the speed limit, it can be
argued that at some level we are well aware of the speed we are travelling. If
that were not the case the presence of a police vehicle would not have such an
immediate impact on traffic speeds. High violators are choosing to drive in a way
that is unacceptable to society, and the way to reduce the level of violating on
the roads is to change the attitudes of the violators. Put simply, we need to find
ways to persuade them not to do it. The threat of sanctions, in the form of penalty
points, fines or a driving ban, are powerful persuaders. Monitoring in the form of
increased police presence and speed cameras also play an important part. However,
the level of enforcement that would be necessary to eliminate the commission of
driving violations would be practically impossible as well as politically unpopular.
It is my contention that stable and enduring reductions in the level of violating
are more likely to be realised through attitude change. While attempting to change
the attitudes of existing drivers is a difficult process, fostering desirable attitudes
among young people who do not yet have a driving licence may be more fruitful
in the long term. Unfortunately, at least in England, to my knowledge little or no
effort is made to impress upon young people that being a good driver is about
being a responsible and courteous driver as well as having the right skills.
Emphasising the predictive link between violations and crashes does not imply
that errors are not important in crash causation. It is obviously necessary for all
 15. Driver Error and Crashes 271

drivers to have the basic level of driving skill that training of novices provides. But it
is a truism that to err is human. Every driver makes mistakes behind the wheel, and
usually, thanks to good fortune and the efforts of engineers in designing primary
and secondary safety features into vehicles and road environments our errors escape
punishment. However, if a driver makes an error while committing a violation, or
if any other road user in the vicinity does, the consequences are more likely to be
disastrous.
The commission of violations takes the driver closer to the limits of his or her
abilities, into a situation where an error of their own, or of anyone else, is far more
likely to be punished, with dreadful results. Therefore, alongside the continuing
efforts to prevent, and recover from, driver error, there must be continuing research
to understand and explain driver violations.

15.5 Aggressive Violations

Having made the initial distinction, we have continued to refine the Manchester
DBQ. For example, in a study focusing on the behaviour of drivers under 40
(Lawton et al., 1997), factor analysis of the DBQ produced a split in the violations
factor. Inspection of the items loading onto each factor showed that the interperson-
ally aggressive violations included in the scale had separated from what we called
the 'normal' or Highway Code violations. In other words, those violations that
in some sense were committed 'against' a specific other road user (racing, chas-
ing, etc.) appeared to be different from those with no immediately obvious victim
(speeding, drink-driving, etc.). To investigate this distinction further, we produced
a 12-item version of the violations scale, that included six interpersonally aggres-
sive, and six normal violations. Table 15.1 shows the wording of these violation
items.
This scale has since been used in a number of DBQ studies (e.g., Parker et aI.,
2002; Lajunen et al., 2004) and the distinction between the two scales broadly
supported. There is also some suggestion that the interpersonally aggressive vio-
lations consist of two further sub-types. The first sub-type of aggressive violation
is related to anger/hostility and may reflect a general personality characteristic.
Lajunen and Parker's (2001) study of personality aggression, anger while driv-
ing and propensity to commit violations provided some support for the idea that
aggressive driving may be seen as a learned problem-solving strategy.
The second sub-type of aggressive violations is to do with gaining advantage
over other road users and maintaining progress. These violations are more likely to
occur in areas of high traffic density, where the motorist comes across unexpected
hold-ups and delays. In our research (Lajunen et al., 1999), traffic obstructions like
traffic jams or road constructions did not seem to provoke anger among British
drivers, a finding supported by Underwood et aI.'s (1999) diary study. These kinds
of impediments may be so common in today's traffic in Britain that drivers can
expect them to occur and therefore do not become unduly frustrated. Hence, the
frustration-aggression hypothesis (Dollard et al., 1939) might be more applicable
272 Parker

TABLE 15.1. Wording of the extended DBQ violation items.

Violation type Item wording

Normal Disregard the speed limit on a motorway.

Aggressive Drive especially close to the car in front as a signal to its driver to go faster or get
out of the way.
Normal Disregard the speed limit on a residential road.
Normal Overtake a slow driver on the inside.
Aggressive Have an aversion to a particular class of road user and indicate your hostility by
whatever means you can.
Aggressive Become angered by another driver's behaviour, and give chase with the intention
of giving him or her a piece of your mind.
Aggressive Sound your hom to indicate your annoyance to another road user.
Normal Drive even though you realise that you may be over the legal blood-alcohol limit.
Normal Cross a junction knowing that the traffic lights have already turned against you.
Aggressive Pull out of a junction so far that the driver with right of way has to stop and let you
out.
Aggressive Stay in a motorway lane that you know will be closed ahead until the last minute
before forcing your way into the other lane.
Aggressive Get involved in unofficial 'races' with other drivers.

in situations where drivers' goals are dramatically blocked by a sudden and unex-
pected event.

15.6 Anger-Provoking Situations

Analysis of data from over 2500 drivers from Finland, the Netherlands and Britain
identified five broad types of situation that provoke anger and/or aggression in
drivers (Parker et al., 2002). Each of the factors identified formed reliable scales
(alpha reliability coefficients 0.73-0.86). The first situation that typically can gen-
erate anger most frequently was one involving a perceived lack of consideration,
or discourtesy, such as when an oncoming driver fails to dip their headlights. The
second most anger-provoking type of situation was when others are putting you in
danger by their driving style, maybe weaving in and out of lanes of slow traffic.
The typical thought response to this might be 'How come he's allowed to get away
with it?' Interestingly, this was the only type of driving that provoked more anger
among older, than among younger respondents, perhaps because it represents a real
threat to your physical safety. The third type of anger-provoking situation involves
a challenge to your personal competence as a driver. You are on the receiving end
of gestures, verbal abuse or a blast of the hom. This threatens your self-identity
as a better than average driver (and nearly everyone thinks they are a better than
average driver). You feel insulted and challenged. The fourth type of situation oc-
curs when your progress is impeded by another driver's hesitance or sluggishness.
You feel thwarted, impeded and may be tempted to believe that the other driver is
doing it purposely, just to inconvenience you. The fifth type of anger-provoking
driver to emerge reflected impatience on the part of the other driver. It is typified
 15. Driver Error and Crashes 273

by the situation when someone nips in and takes the parking space you had been
waiting for.

15.7 Conclusions
Several aspects of human error have been outlined in this paper. It is clear that
intelligent systems can help in some of types of error, for example by reminding
drivers not to dazzle with their lights, or by warning them in advance about unex-
pected problems on their route and suggesting alternatives to prevent or minimise
delay. However, our studies suggest that the reduction of other types of aberrant
driving, that is, driving violations, will also require attention to the social and mo-
tivational aspects of driving. It is crucial to acknowledge that driving is a motivated
behaviour that involves emotions as well as rational decisions.
When drivers know what to expect, they usually cope relatively well. When
something unexpected crops up, that is going to impede their progress, they are
likely to become stressed, react with anger, and begin to take risks, committing
aggressive violations in order to gain advantage over other road users and main-
tain progress. The provision of information may help to prevent these aggressive
violations. At first it may appear that intelligent systems can have their greatest
impact in error reduction. However, it may be that, provided the social aspects of
driver behaviour are taken into account, such systems have an important role to
play in preventing the commission of at least some types of driving violation.

References
Dollard, 1., Miller, N.E., Doob, L.W., Mowrer, a.H. and Sears, R.R. (1939). Frustration
and aggression. Yale University Press, New Haven, CT.
Johnson, P. and Gill, J. (1993). Management control and organizational behaviour. Paul
Chapman, London.
Tversky, A. and Kahneman, D. (1974). Judgements under uncertainty. Science 185, 1124-
1131.
Lajunen, T. and Parker, D. (2001). Are aggressive people aggressive drivers? A study of the
relationship between self-reported general aggressiveness, driver anger and aggressive
driving. Accident Analysis and Prevention, 33, 243-255.
Lajunen, T., Parker, D. and Summala, H. (1999). Does traffic congestion increase driver
aggression? Transportation Research Part F, 2, 225-236.
Lawton, R. (1998). Not working to rule: Understanding procedural violations at work. Safety
Science, 28, 77-95.
Lawton, R.L., Parker, D., Stradling, S.G. and Manstead, A.S.R. (1997). The role of affect in
predicting social behaviors: The case of road traffic violations. Journal ofApplied Social
Psychology, 27, 1258-1276.
Lajunen, T., Parker, D. and Summala, H. (2004). The Manchester Driver Behaviour Ques-
tionnaire: A cross-cultural study. Accident Analysis and Prevention, 36(2), 231-238.
Parker, D. and Lawton, R. (2003). A psychological contribution to the understanding of
adverse events in healthcare. Quality and Safety in Healthcare, 12,453-457.
274 Parker

Parker, D., Reason, J.T., Manstead, A.S.R. and Stradling, S.G. (1995a). Driving errors,
driving violations and accident involvement. Ergonomics, 38, 1036-1048.
Parker, D., West, R.W., Stradling, S.G. and Manstead, A.S.R. (1995b). Behavioural char-
acteristics and involvement in different types of road traffic accident. Accident Analysis
and Prevention, 27, 571-581.
Parker, D., Lajunen, T. and Summala H. (2002). Anger and aggression among drivers in
three European countries. Accident Analysis and Prevention, 34(2), 229-235.
Rasmussen, J. (1990). Human error and the problem of causality in analysis of accidents.
Philosophical Transactions of the Royal Society B: Biological Sciences, 327,449-460.
Rasmussen, J. and Jensen, A. (1974). Mental procedures in real-life tasks: A case study of
electronic troubleshooting. Ergonomics, 17, 293-307.
Reason, J. (1990). Human error. Open University Press, Cambridge.
Reason, J., Parker, D. and Lawton, R. (1998). The varieties of rule-related behaviour. Journal
of Organisational and Occupational Psychology, 71, 289-304.
Reason, J.T., Manstead, A., Stradling, S., Baxter, J.S., Parker, D. and Kelemen, D. (1989). A
report to the TRRL on Research Contract 9885/35. The social and cognitive determinants
of aberrant driver behaviour. Unpublished report.
Reason, J.T., Manstead, A., Stradling, S., Baxter, James, S. and Campbell, K. (1990). Errors
and violations on the roads: A real distinction? Ergonomics, 33(10/11),1315-1332.
Underwood, G., Chapman, P., Wright, S. and Crundall, D. (1999). Anger while driving.
Transportation Research Part F: Traffic Psychology and Behaviour, 2F( 1), 55-68.
 VI
Control Theory Models of
Driver Behaviour
16
Control Theory Models of the Driver
THOMAS JURGENSOHN

16.1 Introduction

From the viewpoint of modelling theory, modelling human behaviour is one of

the most interesting and challenging issues. In this overview, the term 'model'
refers to something 'formal', which can be calculated or simulated. The models
described in the following reflect behaviour of a human during a special task, the
car driving, formalised by means of control theory.
Control theory is a mathematical discipline, which was developed for describing
technical controllers. As a consequence, the application area of control theory is in
modelling the human as a controller. Interestingly, the car driving task is only to a
small extent a controlling task. Driving a car is much more constituted by observing,
deciding, estimating, etc. However, keeping the car on the street and keeping the
distance to the heading car are typically controlling tasks and therefore they seem
predetermined to be described by means of the mathematics of control theory.
After an introduction of different approaches of modelling the human controlling
behaviour in general, the focus in this paper will be on models of controlling the
vehicle path. In the last part, I will shortly discuss modelling speed control and
distance control.

16.2 Modelling Human Controlling Behaviour

Mathematical modelling and subsequent simulation of the human as a machine
operator are very well investigated and elaborated topics within the field of mod-
elling human action in dynamical environments. Often the term 'operator model'
is used synonymously for such models of the human as a controller.

16.2.1 The Tustin-Model: Linear Part + Remnant

Tustin (1947) is considered to have published the first article describing human
behaviour by means of mathematical models outside the military field. He inves-
tigates the behaviour of an operator controlling the rotational speed of a flight

277
278 Jtirgensohn

u
Operator behaviour

y Target

FIGURE 16.1. Behaviour of an operator controlling the rotational speed of a flight defence
cannon via a hand gear (modified from Tustin, 1947).

defence cannon via a hand gear. This is a typical tracking task where the human
has to compensate a deviation. Tustin observed the behaviour of the controllers
(Fig. 16.1), which is fundamentally different from a technical controller: a jerky
curve with 'flats' (Tustin, 1947, p. 119).
Tustin tries to approximate the human controller behaviour by means of a linear
system. The remaining difference U r between approximation u, and real controller
behaviour U he denotes as the remnant (Fig. 16.2).
Both terms of the operator model together, the linear part H(s) and the remnant
Ur , are not an approximation of human behaviour, but describe it completely.
The idea of Tustin was to find a description of the remnant properties which is
independent from the experiment set-up, for example, in a stochastic sense. For
special cases, this is indeed possible (Elkind and Dorlex, 1963; McRuer and
Krendel, 1974). But mostly the remnant has to be considered as an error signal,
which has no special traits, and which cannot be described by means of simple
mathematical methods. However, the remnant part of the operator's output is often

Human Operator

: Controlled Process
y

R = Reference Input U = Control Variable

E = Error Ur=Remnant
Y = Output Variable

FIGURE 16.2. Control theoretic model of the controlling human in the Laplace domain (from
Tustin, 1947).
 16. Control Theory Models of the Driver 279

small, so the linear part characterises most of the control activity. The models of the
human as a controller, which were set up in the following, still contain the remnant,
but this item is usually neglected. In some applications, remnant models can be
useful. But in general, modelling non-linear parts of human behaviour by assuming
a remnant in terms of stochastics is not appropriate for many applications.
The existence of the remnant in the controller models of the human is an example
of an adoption of paradigms from the technical-physical modelling, which is
not appropriate. The successful method for improving the model quality applied
in the technical sector, that is, adding nonlinear parts to the model modelling
approach (e.g., Boeker et aI., 1986) does not resolve the problem of modelling
human behaviour. Human behaviour is variable to such an extent that the refinement
using nonlinear terms only seems to produce a more accurate model.

16.2.2 Laboratory Research, Stochastic Input,

Quasi-Linear Model
After Tustin it became clear very quickly that human control behaviour cannot
be reproduced at large by means of mathematical and system theoretic methods.
Especially the facts that humans not only behave fundamentally differently with
different plants, but that behaviour also depends on the type of input and that there
are vast inter-individual differences as well as intra-individual variations interfere
with a well-defined model.
For sinusoidal reference input signals it can be observed that the human can
follow this signal without phase shift. Such behaviour cannot be explained within
the frame of a causal linear system only for special cases. Figure 16.3 shows
the tracking behaviour of a test person for reference input signals that show a
certain pattern. Contrary to the assumption that models humans as a controller, the
operator does not evaluate the actual control deviation, but reproduces the reference
pattern by means of a motion sequence which is synchronised at discrete times
with the reference signal. This capability to memorise and reproduce the pattern
or the gestalt in certain behaviour cannot be modelled by means of differential
equations.

Reference Input

Operator

FIGURE 16.3. Tracking behaviour of an operator following reference signals with a pattern
(modified from Vossius, 1964).
280 Jtirgensohn

TABLE 16.1. Meaning of the parameters of

the quasi-linear model.
Reaction time
Neuromuscular delay
Adaptation to the controlled plant

In order to come to practical assertions in spite of the strong dependency of

human behaviour from all kind of environmental conditions, researchers in the
engineering area choose the same method as the psychologists: they go to the
laboratory and reduce both tasks and boundary conditions, until reproducible ex-
periment results allow unambiguous modelling.
The task is reduced to pure tracking on a screen and input signals are constrained
to stationary stochastic signals. This assumption will be dropped later and models
are also set up for deterministic input signals, but limited to few cases such as
step function, ramp or sine wave. From the preparatory work done by Krendel
(1951) and Elkind (1956), this research direction was completed in the 1950s
and summarised by McRuer and Krendel (1959), who formulated the quasi-linear
model (Eq. 1).
K . (TLs + 1)
Quasi-linearModel: R(s) = e- rs (1)
(TIS + 1) (TNs + 1)
The quasi-linear model describes the human by means of a linear second-order
differential equation with dead time. Differences in the task environment and in
operator properties are described by means of five free parameters. r (dead time =
reaction time) and TN (neuromuscular delay time) are considered as physiological
parameters independent from the task. The parameters K, TL und TI (Table 16.1)
depend on the control plant and the type of stochastic input signal. This is the
reason for the prefix 'quasi', as the model is only linear in the considered context.
The dependency on the input value is in contradiction with the property of linearity.
In addition, the term exp (- r s) in the model makes it transcendental, rather than
linear.
For the different combinations, a catalogue set-up to be used for the model could
be established (McRuer and Krendel, 1974).
In a practical application to predict operator behaviour in machine environments
still under development the first step is to determine the dynamical behaviour of
the machine in a model. Together with estimated type of nominal values, the
parameters of the quasi-linear model can be looked up in the catalogue. Then the
operator behaviour can be calculated by solving the resulting differential equation
describing the man-machine interaction. The quasi-linear model is up to now the
most general linear model of the human as a controller.

16.2.3 A Holistic Approach: The Crossover Model

The quasi-linear model is unsatisfactory, because the parameters can only be de-
termined very roughly beforehand. However, already in one of the first systematic
 16. Control Theory Models of the Driver 281

studies which were conducted by Elkind (1956) and Ekind and Forgie (1959)
dependencies between the parameters could be noticed. The operator adapts his
behaviour in such a way that the overall system behaviour remains approximately
constant in spite of different nominal values. A similar assertion is formulated by
McRuer und Krendel (1962) in their crossover-model. Experimentally measured
frequency responses of the total system exhibit a common property under different
conditions. If the open-loop linear transfer function L(s), which describes human
and machine together, is calculated from the observed closed-loop behaviour, then
the behaviour can be very well approximated by an integrator and an additional
phase correction by means of a dead time for all task environments in the region
of the crossover frequency, We (L(jw) == 0 dB). The crossover model requires only
two parameters, that is, the crossover frequency We and the dead time t , which is
not identical with the dead time of the quasi -linear model.
We
Crossover model: L(s) == _e- rs (2)
s
The interesting point of this model is that human and machine are not modelled
separately as successive connected structures, but as one system. The focus is not
on how humans manage to adapt to different environments, but this is assumed to
be the case. Only total system behaviour is considered. It should be noted that the
crossover model is not a rough approximation with very few parameters, but in
many cases a very precise description.
In a certain way the crossover model already contains the element of goal-
oriented action which dominated in recent approaches. The great similarities in
the frequency responses of the man-machine system for various types of control
plants are manifestations of the operator will. The operator wants to minimise the
control error and achieve a certain overall system behaviour, a goal that is indeed
accomplished for very different types of control plants. The behaviour of the
operator is different, but the overall system behaviour is approximately identical.
Although the crossover model contains internal goals of the operator implicitly
and is therefore more suitable to predict human control behaviour within unknown
control plants than the quasi -linear model, the parameters of the crossover model
depend on the type of the input function, too. Therefore, the description of the
influence of boundary conditions on goal achievement is still only possible a
posteriori or via a catalogue. McRuer and Krendel (1974) and other authors es-
tablished comprehensive collections of crossover model parameters for different
system environments and nominal signal types. As these parameters are not re-
lated to other invariant human properties, the validity of the crossover model for
predicting operator behaviour in unknown environments is also limited.

16.2.4 Nonlinear Approaches: Improved Reproduction

ofMeasured Behaviour
The crossover model is the most compact quantitative description of human dy-
namical behaviour. But it is still a linear model and has - as the quasi-linear
282 Jtirgensohn

0 -.............-+-.......--+-................---.....- ......- -.+...........--.--+-..........................

t [sec]
~

FIGURE 16.4. (Top) Control action of a double integrating control plant by a very well-
trained pilot (from McRuer et aI., 1968). (Bottom) Steering motion of a helmsman on a ship
(from Veldhuyzen and Stassen, 1976).

model- only limited capabilities to describe non-stationary behaviour. Non-

stationary behaviour is distinguished by discontinuities with a change between
action and hold phases (Fig. 16.4).
In the 1960s and 1970s, diversification of the approaches and experimenting
with the evolving methods of control theoretic was prevailing in the research work
within the field 'human controller': the models developed included sample and
hold models for discontinuous behaviour, bang-bang controllers for time optimal
behaviour, state space models for multidimensional sensors and Kalman filters for
disturbed estimation processes (extensively described in Johannsen, 1980). Due to
the great part of special system theory mathematics, the operator modelling was
almost entirely advanced by engineers.
Due to space restrictions, it is impossible to go into details of all the many
approaches that were put forward. It is important that the different models can re-
produce human behaviour for the investigated experiments, but that the prediction
quality for other boundary conditions or other tasks is quite poor. The paradigm
of describing dynamic behaviour by means of differential equations remains un-
questioned. The main issue is just to find the right differential equation, be it linear
or nonlinear.
Similar to the psychological theories of that time, who - while competing for
truth - were caught in their specific framework of possible explanations in the sense
of a falsification process according to Popper and were only able to explain certain
aspects in a satisfactory way, the applicability of the different model approaches
of the human controller were limited to the respective aspects of behaviour.
The recourse to the complex control theoretic method stock was based on the
hope to model complex human behaviour better than by means of simple linear
 16. Control Theory Models of the Driver 283

differential equations. On the other hand, the choice of the method is often also
due to the attraction of a 'modern' tool. Conceptional reasons are often given
only ex post. An example is the BBN model by Baron and Kleinman (1969),
which is a linear model based on an optimal filter approach. Retrospectively, this
model was often praised as especially appropriate, because it contains elements,
which can be interpreted as 'internal model'. In another example the advantage of
models containing adaptive control approaches is deduced from the adaptivity of
human behaviour. The suitability of modelling approaches thus is justified with
resemblances of the model to observed characteristics of human behaviour.

16.3 Driver Models for Vehicle Design

For the majority of automotive vehicle engineers the term driver model means a
dynamic driver model, combined with models of lateral vehicle dynamics. These
driver models were developed to obtain information on how a design change influ-
ences the handling quality. Allen (1983) has called this kind of driver model'design
driver'. As the vehicle behaviour is generally described by means of differential
equations, driver models are usually formulated in this way, too. Especially for
stability investigation of the complete system, it is useful to employ the method of
linear control theory. Although there are alternative models, up to now the models
using differential equations are most popular. Not until the seventies algorithmic
models, models with decision-making structures, models using fuzzy decision or
fuzzy control or artificial neural networks have been established.
Linear driver models have already been reviewed in detail (see, e.g., Hoffmann,
1975; Billing, 1977; Reid, 1983; Reichelt, 1990; Guo and Guan, 1993; Willumeit
and Jurgensohn, 1997; Jurgensohn, 1997). Most driver models developed hereto-
fore model the driver as a controller of lateral vehicle dynamics. These models
are adapted from the so-called tracking-models in which the human manipulates a
machine according to an externally set value (Fig. 16.5). The prevalent descriptions
that are used are linear differential equations or difference equations.
These driver models were developed to anticipate influences of design on the
handling quality. As the vehicle behaviour is generally described by means of
differential equations, driver models are also usually formulated in this way. Espe-
cially for stability analysis of the complete system, it seems to be useful to employ
methods of linear control theory.

Driver
Vehicle
Set/Desired Value Controlled Value
y
L , J _---/

V
L(s)

FIGURE 16.5. The driver as a controller.

284 Jtirgensohn

FIGURE 16.6. 'Shaft model' of driver's control strategy. Copyright, Jurgensohn (2000).

The lateral control driver models were initiated in Japan. In 1953, Kondo (1953)
developed two different models for driver behaviour due to side wind disturbance
and calculated the behaviour of the whole system on a straight path with constant
speed by means of a single-track model. Both models already show the general
structure of the majority of models to be published for the following 35 years.
In the first model, Kondo assumes that the driver always steers in a way that an
imaginary point within a certain preview distance L, the sight point or aim point,
is on a predetermined course (Fig. 16.6). In terms of control theory this would
mean that the driver minimises the lateral offset ~ y L of the vehicle's projected
centreline from the desired course within the preview distance L in front of the
vehicle.
His assumption was inspired by a situation after an accident in which his car was
towed with a very short rope. For this reason the model is often called the shaft
model. He was also inspired by an experience by driving bicycle in the narrow
streets of Japanese cities: In a narrow lane it is easier to hold the bicycle steady
looking to the exit of the lane. This linear prediction model has been adopted by
many authors, especially in Japan and later in Germany (Fiala, 1967).
The model depicted in Fig. 16.6 can be described in control theoretic terms in
three different ways. First of all ~YL can be taken as a prediction of ~Yawith a
prediction time Tp For a simple P -controller it follows that

(3)

where Tp == L/va are the preview time and Va the longitudinal speed.
Another control theoretic formulation of the model of Fig. 16.6 can be found
for small values of the sight angle <t>.

8H(s) == K . (~Ya(s) + L . ~1/J(s)) (4)

 16. Control Theory Models of the Driver 285

Driver Vehicle
,----11-----------;-----1
I
I
s, II bT(sE-At lT
X
C ~-~
I Llyo
1++ I
I ______ 1
I
I c=[l,O,O]T
I x T=[LlyO,Llv,Ll'V]
I
L=[k 1,kbk3]T
I I
I I
~ J

FIGURE 16.7. Model of driver's control strategy as state-variable controller (Kondo, 1953).

With this transformation the model is reduced to a control of the centre of gravity.
For small steering angles, we can write:

(5)

and the following simple preview predictor model results:

(6)
The latter equation is in fact a linearisation ofEq. (3): A lead term as an approx-
imation of a time-preview term.
Kondo's second model from 1953 assumes that the steering angle is a linear
combination of yaw angle error (~ 1/1), heading angle error (~ v) and lateral position
error in the gravity centre (~Yo). In control theoretic manner this can be described
as a linear state variable controller (Fig. 16.7). This second model is modified and
used in many other models, too (e.g., Donges, 1978; Reid et al., 1981).
In another kind of graphical representation, this model has the following form
(Fig. 16.8).
Another well-known model is the so-called 'STI' -model. Systems Technology
Inc. dominated the research in the USA with already mentioned scientists such
as McRuer, Weir, Klein and others. They developed a model (Fig. 16.9), which
was derived from tracking models and pilot models. Besides the model class of
preview predictor models and linear state control models invented by Kondo, the
STI model can be seen as another model class.
A 'translation' of this cybernetic view of the model into a control theoretic
version (Fig. 16.10) shows the similarity to the model of Kondo.

FIGURE 16.8. Linear state control model of the driver

(Kondo, 1953).
 286 Jtirgensohn

i
DISTURBANCES
~RIVER - - - - - - - DR~E;j ( ~IND8ROAD ) ~AR/ROAD -;;OTlO;KINEM;:I~- ;
1 REMNANT I d I I
,..- ..," I V t:.y
I r I
I r5 8$ CAR LANE I
(TL S + l)e- I DYNAMICS I DEVIATION
I rI I
I : I HEADI ~G I
I I ROAD I U ANGL E :
IL DRIVER PERCEPTION
- J
I CURVAl\JRE p'-I
r L DEVIATION I
..J

VISUAL SCENE I
SIMULAT ION DISPLAY REL ATIVE HEADING ANGLE ~o/

RELATIVE L ANE DEVIATION t:.y

C Jurgensohn,1999

FIG URE 16.9. 'STI' -model (McRuer et a!., 1977).

A noteworthy difference to the approach of Kondo is the existence of the time-

delay term T. The purpose of this term is to reflect the reaction time of the driver
(usually a few seconds). We know this term from many models published in the
early 1950s to represent human behaviour when controlling an aircraft. In contrast
to car driving, the control task in an aircraft was, in the 50s, dominated by com-
pensating disturbances. Car driving, in opposite, is more dominated by following
a curved band, which can be seen far ahead. So, during normal driving conditions,
the driver has no reaction time with respect to the task 'following the street' . The
STI model was derived from the classical models of human compen satory be-
haviour, whilst Kondo started as a mechanical engineer from the stretch without
preconceived notion.
Besides these two main concept s we can find a third basic idea in many control
theoretic driver models : the concept of anticipatory open-loop control. According
to this approach, the driver does not react to variables of the vehicle only, but to

FI GURE 16.10. Block diagram of the 'STI'

model..
 16. Control Theory Models of the Driver 287

desired
path curvature .... V002eTA
(82+2<;00'8+002)
I11III""""

FIGURE 16.11. Two-level model of Donges (1978).

street variables, too. This type of driver model was published first by Fiala (1967)
and Ohno (1966). It assumes that part of the steering angle is determined from
lane curvature K (see Fig. 16.11) at a particular heading distance. This is open-loop
control, because the lane curvature is independent from vehicle motion. Fiala called
this 'Scheinwerferorientierung' (head light orientation), whereas Ohno talks about
programmed action and intuitive steering. In Fig. 16.10, the structure of one of the
most renowned models with an anticipatory open-loop part, the 'two-level model'
by Donges (1978) is shown. It is a combination of the state variable controller in
the closed-loop as shown in Fig. 16.7 and an open-loop term, assumed as a linear
system with time prediction having 2 df. The reaction time i H of the driver is
modelled by a time delay term. The remnant is an additional source of unknown
influences (driver noise) - also adopted from the old pilot models.
The idea of a compensatory closed-loop control model with an additional open
loop part has found many imitators (Horn, 1986; Mitschke, 1977; Reid et al., 1980).
Yet another step further ahead are the models by Bosch (1991) and Plochl and
Lugner (1994), who introduce a third additive scope of control, the so-called 'local
control' (correction of the instantaneous path deviation) besides compensatory
control and anticipatory control.
Moreover, there are models, which are capable to change their structure during
simulation, for example the dual mode model (Me Ruer et al., 1977) which can
switch between open-loop and closed-loop behaviour. This model has been realised
in DRIVEM (Lieberman and Goldblatt, 1981). This kind of modelling can also
be found in adaptive models which link sudden changes in the environment, for
example, sudden appearance of black ice (Reichelt, 1990; Nagai, 1983; Nagai and
Mitschke, 1987), with changes in the model.
Besides these four types of driver models there are some variations, which are
distinguished by dead times, the kind of input or output parameters or by non-
linear elements. For example Baxter and Harrison (1979) include non-linearities
288 Jtirgensohn

by assuming a hysteresis or Carson et al. (1978) include perception thresholds.

Some models are formulated in a time discrete form with sample and hold terms
(Crossman et al. 1966; Hayhoe, 1979; Reid et aI., 1980). MacAdam (1980) and
Blaauw et al. (1984) determined the parameters of their model by an optimisation
criterion. Included in this class of preview models are also models, which choose
the sight angle <P (Fig. 16.6) as an input parameter (Baxter and Harrison, 1979;
Mitschke, 1977).
For most models the driver output is the steering angle 8H. There are models,
however, which take the steering wheel rate (Hayhoe, 1979) or the steering torque
(Fiala, 1967) as output of the driver. The desired input is usually a future lateral
deviation or a path curvature. Very often still other state variables such as roll
angle (Niemann, 1972) are used as input. Reichelt (1990) can identify 21 different
information parameters for the driver, from attitude angle rate (Braess, 1970) to
the third derivative of the reference curvature (Fiala, 1967).
These early examples of driver models briefly reviewed here represent the ma-
jority of all known models of driver's lateral control activity. Their structure is very
similar to mechanical systems plus a controller applied to such systems. The advan-
tage of this modelling method is, as mentioned before, that we are able to describe
and examine the whole system with a well-known mathematical apparatus.
An example for a commercial driver model currently used in real-time vehicle
dynamics simulations is the ve-DYNA advanced driver controller. This model
guides the vehicle along a given trajectory, especially near the driving limits at
high speeds, for example, for race line optimisation. Other application fields are
virtual driving tests like double lane change for handling investigation, comfort
studies, as well as various tests for electronic control units like ESP, active body
control, etc. (software-in-the-loop and hardware-in-the-loop),
Most electronic vehicle controllers are designed to support the driver in master-
ing difficult driving manoeuvres, typically caused by driving errors. To test these
support functions, critical driving situations have to be realised, for example, driv-
ing through curves at high speed near the driving limits. For reasonable testing of
controller behaviour in the vehicle environment, it is therefore desirable that the
virtual driver, that is, the driver model also exhibits non-perfect driving behaviour,
like ordinary human drivers do.
The driver model is therefore realised as a technical controller considering
some human characteristics. Different from technical controllers, a human driver
does not simply follow a given trajectory, but sets the target course within given
constraints (i.e., road width or lane width). This is reflected in a two-level structure
of the driver model, consisting of guidance and stabilisation (see Fig. 16.12).
At the guidance level, the time-dependent target position, target speed, target
direction and target path curvature are determined. These values serve as nominal
values for the stabilisation level which is basically a position controller based on
the theory of nonlinear system decoupling and control supplemented by a model
for perception of vehicle state information (e.g., side slip angle).
The position controller at the stabilisation level keeps the vehicle as near as
possible to the target point moving along the given track, as illustrated in Fig 16.13.
 16. Control Theory Models of the Driver 289

Maneuver
Definition Vehicle State Vehicle
Dynamics

target Path
.... ... Stabilisation Level ...._ .....
(Position Control)

Feedback Control Loop

Feedback to Guidance
(e.g. adaptation to enginepower)

FIGURE 16.12. Structure of the advanced driver (TESIS GmbH, Munchen).

-- ~Vehicle X(t), Y(t), v(t)

FIGURE 16.13. Target trajectory tracking and position control.

The distance between target point and vehicle is not fixed. One could imagine an
elastic linkage between the vehicle and the moving target point towing the vehicle:
The more the vehicle deviates from the target, the harder it will be pulled towards
the target. In the ideal case the driver will keep the vehicle just atop the target at
any time.
Individual driving behaviour like differences in risk-taking and driving expertise,
as well as typical driver errors can be reproduced by means of parameters of
the driver controller, for example, the preview distance, delay times for steering
response, etc., and by influencing the target course (Irmscher and Ehmann, 2004).

16.4 Summary and Future Prospects

This short presentation of examples for models of the car driver shows that the
control theoretic core of the models has not changed much from the beginnings in
the 1950s up to now. Although much time was spent for iterative improvements,
290 Jtirgensohn

there was just a small, 'model evolution' as in many other fields of modelling.
The reason can be found in the large inter- und intra-individual differences and
deviations of human dynamic behaviour. A precise and differentiated modelling
of dynamic human behaviour is not possible or not meaningful. In general, it does
not make sense to describe driver behaviour by means of differential equations of
higher than second order. In a few applications, well-established neuromuscular
models for arm and hand actuation can add an additional third order. Such models
can be useful, for example, in studying steering characteristics.
Therefore, the focus of driver model research shifted from control theory to
other modelling methods, which are better suited to describe non-dynamic aspects
of behaviour, for example, decision making or planning. However, the last example
(cf. Fig. 16.13) illustrated that for the description of dynamic steering characteris-
tics of the driver control theory is still the most appropriate means. However, the
incorporation into an algorithmic environment is the central point here. Summing
up, it can be said sufficient research on driver models based on control theory has
been done, but that there are still open issues on the problem of its integration into
a 'hybrid modelling' (Jiirgensohn, 1997).

References
Allen, R.W. (1983). Defining the design driver from a vehicle control point-of-view. 62nd
Annual TRB Meeting, Washington, USA, STI-Report, No. 83327, S. 1-11.
Baron, S. and Kleinman, D.L. (1969). The human as an optimal controller and information
processor. IEEE Transactions on Man-Machine Systems, Vol. MMS-I0(1), 9-17.
Baxter, 1. and Harrison, 1.Y. (1979). A nonlinear model describing driver behaviour on
straight roads. Human Factors, 21(1), 87-97.
Billing, A.M. (1977). Modelling driver steering in normal and severe manoeuvres. In J.E.
Bernard (Ed.), An overview of simulation in highway transportation, Part II. Bosten:
Simulation Councels, Inc. (S. 151-165).
Blaauw, GJ., Godthelp, H. and Milgram, P. (1984). Optimal control model applications and
field measurements with respect to car driving. Vehicle System Dynamics, 13,93-111.
Boeker, J., Hartmann, I. and Zwanzig, Ch. (1986). Nichtlineare und adaptive Regelsysteme.
Springer-Verlag, Berlin.
Bosch, P. (1991). Der Fahrer als Regler (the driver considered as a control element). Dis-
sertation, TV, Wien.
Braess, H.-H. (1970). Untersuchung des Seitenwindverhaltens des Systems Fahrzeug -
Fahrer (investigation of side gust behaviour of the driver-vehicle system). Deutsche
Kraftfahrtforschung und Straj3enverkehrstechnik, Heft. 206, 5-32.
Carson, 1.M., Wierwille, W.W. and Eastman, C. (1978). Development of a strategy model
of the driver in lane keeping. Vehicle System Dynamics, 7(4),233-253.
Crossman, E.R.F.W., Szostak, H. and Cesa, T.L. (1966). Steering performance of automobile
drivers in real and contact-analog simulated tasks. Human Factors Society 10th Annual
Meeting. Anaheim, California.
Donges, E. (1978). A two-level model of driver steering behaviour. Human Factors, 20(6),
691-707.
Elkind, J.1. (1956). Characteristics ofsimple manual control systems. M.I.T. Lincoln Lab-
oratories, Lexington, Massachusetts., Technical Report, No. 111.
 16. Control Theory Models of the Driver 291

Elkind, 1.1. and Forgie, C.D. (1959). Characteristics of the human operator in simple manual
control systems. IRE Transactions on Automatic Control, 4(1), 44-55.
Elkind, 1.1. and Dorlex, D.L. (1963). The normality of signals and describing function
measurement of simple manual control systems. IEEE Transactions on Human Factors
in Electronics, 52-55.
Fiala, E. (1967). Die Wechselwirkung zwischen Fahrzeug und Fahrer (Interaction between
vehicle and driver). AutomobiltechnischeZeitschrift, 69(10), 345-348.
Guo, K. and Guan, H. (1993). Modelling of driver/vehicle directional control system. Vehicle
System Dynamics, 22, 141-184.
Hayhoe, G.E. (1979). A driver model based on the cerebellar model articulation controller.
Vehicle System Dynamics, 8, 49-72.
Hoffmann, E.R. (1975). Human control of road vehicles. Vehicle System Dynamics, 5,
105-126.
Horn, A. (1986). Fahrer-Fahrzeug-Kurvenfahrt auftrockener Strafie (driver-vehicle curve
driving on dry road). Dissertation, TV, Braunschweig.
Irmscher, M. and Ehmann, M. (2004). Driver Classification using ve-DYNA Advanced
Driver, SAE paper 2004-1-0451.
Johannsen, G. (1980). Manuelle Regelung in Mensch-Maschine-Systemen. Habilitation
RWTH Aachen.
Jurgensohn, T. (1997). Hybride Fahrermodelle (Hybrid driver models, ZMMS-Spektrum,
Band 4). Sinzheim: Pro Universitate Verlag.
Kondo, M. (1953). Directional stability (when steering is added). Journal of the Society
ofAutomotive Engineers ofJapan (JSAE), Jidosha-gijutsu, 7(5/6) (in Japanese). Tokyo,
104-106, 109, 123, 136-140.
Krendel, E.S. (1951). A preliminarystudy ofthepower-spectrumapproachto the analysis of
perceptual-motor performance. Wright Air Development Center, Wright-Patterson Air
Force Base, Dayton, Ohio, Technical Report, No. 6723.
Lieberman, E.B. and Goldblatt, R.A. (1981). A review of the driver-vehicle effective-
ness model. Final Report under NHTSA Contract,DTNH22-80-C-07082 (DOT-HS
806110).
MacAdam, C.C. (1980). An optimal preview control for linear systems. Journal ofDynamic
Systems, Measurement, and Control, 188-190.
McRuer, D.T. and Krendel, E.S. (1959). The human operator as a servo system element
Part I + II. Journal of The Franklin Institute, 267(5/6), 381-403, 511-536.
McRuer, D.T. and Krendel, E.S. (1962). The man-machine concept. Proceedings of the
IRE, 50, 1117-1123.
McRuer, D.T., Hofmann, L.G., Jex, H.R., Moore, G.P., Phatak, A.V., Weir, D.H. and
Wolkovitch,1. (1968). New Approaches to Human-Pilot/vehicle Dynamic Analysis. Air
Force Flight Dynamics Laboratory, Technical Report, AFFDL-TR-67-150.
McRuer, D.T. and Krendel, E.S. (1974). Mathematical models of human pilot behavior.
Agardograph, No. 188. Technical Editing and Reproduction Ltd., Harford House, Lon-
don.
McRuer, D.T., Allen, W.R., Weir, D.H. and Klein, R.H. (1977). New results in driver steering
control models. Human Factors, 19(4),381-397.
Mitschke, M. (1977). Kraftfahrzeug-Fahrer-Aktive Sicherheit (Vehicle-driver-active
safety). Automobil-Industrie, 3, 29-32.
Nagai, M. (1983). Adaptive behaviours of driver-car systems in critical situations. Pro-
ceedings of the International AMSE Summer Conference "Modelling and Simulation".
Nice, France.
292 Jurgensohn

Nagai, M. and Mitschke, M. (1987). An adaptive control model of a car-driver and computer
simulation of the closed-loop system. In M. Apetaur (Ed.), The Dynamics of Vehicles on
Roads and on Tracks. pp. 275-286.
Niemann, K. (1972). Messungen und Berechnungen iiber das Regelverhalten von Auto-
fahrern (measurements and calculations of the control behaviour of drivers). Dissertation,
TU, Braunschweig.
Ohno, T. (1966). Steering control on a curved course. Journal of the Society ofAutomotive
Engineers of Japan (JSAE), Jidosha-gijutsu, 20(5) (in Japanese), 413-419.
Plochl, M. and Lugner, P. (1994). Theoretical investigations of the interaction driver-
Feedback-Controlled automobile. Proceedings of the International Symposium on Ad-
vanced Vehicle Control 1994., SAE Paper, No. 9438006, pp. 42-48.
Reichelt, W. (1990). Ein adaptives Fahrermodell zur Bewertung der Fahrdynamik von
PKW in kritischen Situationen (An adaptive driver model to evaluate vehicle dynamics
in critical situations). Dissertation, TU, Braunschweig.
Reid, L.D. (1983). A survey of recent driver steering behaviour models suited to accident
studies. Accident Analysis and Prevention, 15(1), 23-40.
Reid, L.D., Graf, W.O. and Billing, A.M. (1980). The validation of a linear driver model.
University of Toronto, UTIAS Report, No. 245.
Reid, L.D., Solowka, E.N. and Billing, A.M. (1981). A systematic study of driver steering
behaviour. Ergonomics, 24(6),447-462.
Tustin, A. (1947). The nature of the operator's response in manual control and its implica-
tions for controller design. IEEE Journal, 94, Part II A2, 190-202.
Veldhuyzen, W. and Stassen, H.G. (1976). The internal model. What does it mean in human
control? In T.B. Sheridan and G. Johannsen (Eds.), Proceedings ofMonitoring Behavior
and Supervisory Control., Berchtesgaden, Germany, pp. 157-170.
Vossius, G. (1964). Die Vorhersageeigenschaften des Systems der Willkurbewegung,
Neuere Ergebnisse der Kybernetik. Munchen, Wien: Oldenbourg Verlag, pp. 196-209.
Willumeit, H.-P. and Jurgensohn, T. (1997). Driver-models: A critical survey. ATZ
Worldwide, Vol. 99, Part 1 + 2 in No.7 + 8. Wiesbaden: Vieweg, pp. 19-22/19-25.
17
Review of Control Theory Models for
Directional and Speed Control
DAVID H. WEIR, PhD AND KEVIN C. CHAO, MS

17.1 Introduction
Models of driver steering control in regulation tasks are well established and have
been used in a number of studies of driver/vehicle response and performance, e.g.
Weir (1973), Weir and McRuer (1968, 1973), Weir and DiMarco (1978), Weir et al.
(1982) and McRuer et al. (1975, 1977). There is a large literature on this general
topic, including the work of Hoffman (1975), Donges (1978), Sheridan and Ferrel
(1981), Godthelp (1985), Levison and Cramer (1995), Allen et al. (1996) and the
recent survey article (MacAdam, 2003), among many others.
The directional control models of interest here apply to straight or curving
roadways, with approximately constant speed operation and steering to stay in
the centre of the lane in the presence of a random (directional) yaw disturbance.
The models can involve multiple loops or perceptual feedbacks. They are typically
expressed in Laplace transform (transfer function) or differential equation form,
in a classical control theory manner.
The driver model for speed control is similar in form to the model for direc-
tional control, but simpler. The model of interest here is for speed regulation,
where the driver uses the accelerator pedal to maintain a specified constant speed
in the presence of a perceived speed error. The speed error could result from a dis-
turbance such as a headwind gust or could arise in a car-following situation. The
more general topic of car following can involve a number of additional factors not
addressed here, including nonlinear and deterministic behaviour, e.g. Brackstone
and McDonald (1999), Brown et al. (2001) and Fancher and Baraket (1998). The
speed regulation model described herein is a single-loop model. An additional
driver output, and perhaps additional feedback loops, can be added for braking
control.
Useful and simple forms of the driver control models are based on the'crossover
model' of the human operator, e.g. McRuer and Weir (1969) and McRuer and
Krendel (1974). This model applies to a wide variety of manual control tasks,
including driving a car or piloting an aircraft. It can be used to model single-
loop activity such as speed control. It can also model the inner and outer loop

293
294 Weir and Chao

control activity in multiple-loop tasks, as illustrated subsequently. The crossover

model quantifies driver control behaviour, based on task variables, such as the
vehicle dynamics and the disturbance and command inputs to the driver/vehicle
system. It is both descriptive and predictive of driver control activity. Given a
quantitative description of the task variables, it can be used to estimate and predict
driver control behaviour by use of a describing function (transfer function-like,
input/output relation).
The driver models can be quantified for a particular task using a driving simu-
lator or a suitably equipped instrumented vehicle. A driving simulator, computer
simulation, or analytical methods can be used to apply the models, e.g., in the
analysis or optimisation of a vehicle control subsystem.

17.2 Basic Crossover Model of the Human Operator

The crossover model of the human operator was originally developed and applied
in the 1960s (see McRuer and Krendel, 1974). This quasilinear model and its
extensions have been used, subsequently in a wide variety of studies involving
the prediction, analysis, and measurement of aircraft and ground vehicle control
and performance. The crossover model underlies the more specific driver control
models and embodies important manual control principles. It is a useful place to
start.
The crossover model can be described using the simple single-loop system of
Fig. 17.1. The system input (i) can be either a command to follow or a disturbance.
The operator is the driver, and the controlled element is the vehicle dynamics, in
the driving task.
The driver describing function has the form

y; == K e
-TS (TLs + 1) (1)
P P (TIS + 1)
where K p is the gain, adopted by the driver, T is the driver's time delay, TL is a
lead element reflecting anticipation, TI is a lag element reflecting smoothing and
S is the Laplace transform variable.
These parameters are adjusted according to the model, and the values of TL
and TI depend on the vehicle dynamics. The quasilinear model, Yp is called a
describing function, because it characterises only the linear part of the driver's

e C m

.
~-
Yp ---
....- Yc .....
OPERATOR CONTROLLED
ELEMENT FIGURE 17.1. Single-loop
manual control system.
 17. Review of Control Theory Models for Directional and Speed Control 295

input/output or control activity. The nonlinear part is called remnant, and it can
be modelled as an additive noise (McRuer and Krendel, 1974). The linear part
is the main topic here, and it typically accounts for most of the driver's control
activity.
The model is applied by considering the open loop-properties of the
driver/vehicle system (YpYe ) . The crossover model says that the driver adjusts
the lead and lag terms (TL and TI ) as necessary to achieve a K/ s-like amplitude ra-
tio characteristic in the frequency region of closed-loop control. This is the region
of the crossover frequency (we), where the closed-loop 0 dB value occurs, based
on the driver's gain (K p ) . For example, if the vehicle dynamics are Ke/s, then no
lead or lag equalisation is needed, and the driver describing function is simply a
gain plus a time delay. This is illustrated in the Bode (frequency response) plot of
Fig. 17.2, which shows the amplitude and phase properties of YpYe for Ye = K e / S,
TL = TI = 0, and a time delay of 0.3 s. The amplitude ratio in Fig. 17.2 is seen
to represent a K/s transfer characteristic (a slope of -20 dB/decade). The phase

20

~
C 10
"(ij
C> odB (closedloop)

-10

-20 L--------'L....-.............. --'--.L-L-J.....L..J..----I-.L...---'---"--"'--'--J-J,...I"...L--_'---~'___'__~~

10.1

-90r- _

-180

-270'-----_'----'----''___'__~....L..L__----'----'---.L...-..L-..I.-...............'''''---_'''---_'__'___'_...............~
10.1
Frequency (radlsec)

FIGURE 17.2. Open-loop driver/vehicle system bode plot.

296 Weir and Chao

angle plot shows 90 of phase lag at low frequency, corresponding to K/ s. There

is additional phase lag at mid and high frequency due to the driver's time de-
lay (r). Typical values of time delay in the continuous control tasks for which
this model applies are typically in the range of 0.2 to 0.3 s and the values de-
pend on the controlled element dynamics, input bandwidth, manipulator char-
acteristics and driver factors such as skill and level of attention, among other
things.
It should be noted that time delay values of the order of 0.3 s reported in the
manual control literature often resulted from experiments with relatively simple
and well-defined displays, lightweight manipulators and reasonably skilled sub-
jects (engineers or pilots). As will be shown subsequently, more mature typical
drivers using automobile controls can have larger values of time delay and cor-
respondingly lower crossover frequencies. The phase margin values are similar,
however, reflecting desirable damping and stability qualities.
The crossover model can be used to quantify and analyse such things as
driver/vehicle system closed-loop stability, on-centre control characteristics, di-
rectional and speed disturbance regulation, path following and easy manoeuvres,
the terminal control (residual error correcting) phases of abrupt manoeuvres, and
driver behaviour with vehicles whose dynamics are augmented by a stability con-
trol system. As illustrated subsequently, the crossover model can be applied to a
single-loop system, or to successive inner- and outer-loop closures of a multiple-
loop system such as the driver/vehicle model for path control.
From a control standpoint, when an open-loop system has the properties shown
in Fig. 17.2, it is a good system. The large low frequency amplitude ratio provides
good error regulation. A comparatively high closed-loop bandwidth (w c ) gives
good performance in following the command input (i) and helps error regulation.
The phase margin (phase angle at wc) is a measure of the closed-loop stability and
damping, which can be set at a desired level by adjusting the operator's closed-loop
gain (0 dB line).

17.3 Model for Driver Steering Control

Figure 17.3 shows a model structure for steering control (terms used in the figure
are defined below). The model contains the following:

An inner feedback loop of heading angle (1/1). This loop equalises the outer loop,
and makes it easy for the driver to control lane position. It also provides path
damping.
An outer feedback loop of lateral lane position (y). This provides for lateral
deviation control and causes the driver/vehicle system to follow the desired path
input (Yc).
A random appearing yaw rate disturbance input (rd) that can be used to help
define the driving task and as a basis for identifying the driver describing function
components.
 17. Review of Control Theory Models for Directional and Speed Control 297

Yaw
Disturbance
rd
!DRivER------------- ---- --- ---------[ :
Roadwaypath : Yc Ye :0: Actual
~~ :: roadway path
'-------'
~ ___ _ ~ ~ J
Headingangle
Lane position

FIGURE 17.3. Model for directional control with steering.

An alternative and nearly equivalent inner loop candidate is path angle, which
can be used if that is preferred. Path angle involves perceiving the total velocity
vector or focus of expansion relative to the surround, whereas heading angle relates
a position vector (line) in the vehicle to a geometric reference in the surround (e.g. a
lane centreline or edge line). Other possible feedbacks such as sideslip and lateral
acceleration are less good from a perceptual standpoint and less good in terms
of desirable manual control qualities (path following, path damping, etc.). These
feedbacks may playa supplementary role in some manoeuvres, such as near limit
turns, which the model in Fig. 17.3 does not address.
Referring to Fig. 17.3, the dynamic model of the vehicle (YI) is known or can be
measured. The driver describing functions Yy and Y1/1 can be predicted or estimated
using a combination of the crossover model, control principles and empirical results
from driving simulator or instrumented vehicle studies. The describing functions
can also be measured using a driving simulator or test vehicle, and correlation
or spectral analysis methods. The main driver model parameters are gains and
a time delay (r). Equalisation (lead or lag, anticipation or smoothing) can also
be included, but this refinement is often not needed for contemporary passenger
vehicles.
The specific forms of the driver describing function Y1/1 and Yy can be estimated
for an example typical passenger car by successive application of the crossover
model. Consider the (inner) heading angle loop. The vehicle dynamics can be given
by the transfer function relating yaw rate to steer angle which can be approximated
by
yr _ Kr Uo/(a+b)
(2)
8 - (T; + 1) (T, + 1)
Here, T, is the vehicle's yaw rate time constant, which typically has a value of
about 0.1 s for modern passenger cars at a typical highway speed. U0 is the speed
and a + b is the wheelbase. So, in the region of crossover for the inner loop in
Fig. 17.3 (about 1 to 2 rad/s), YI is approximately a gain and y81/1 is approximately
K/s (heading angle is the integral of yaw rate). Therefore, applying the crossover
model, it follows that the inner loop driver model in this case is

(3)
298 Weir and Chao

and Y1/! Yo1/! has the form of Fig. 17.2, where

1/! 1 r
Yo = -Yo (4)
s
Choosing a typical value of time delay, t = 0.3 s, and a driver gain (K1/!) of about
0.25 front wheel angle/" heading error, gives an estimated crossover frequency
of about 2 rad/s and a phase margin of about 40 for good closed-loop damping.
The resulting open outer loop driver/vehicle transfer function is approximately

Y Yy u, (5)
Ye (~s + 1) (T,s + 1) s

and for a 10ft wheelbase and a speed of 80 ft/s

~ _ y 80
(6)
Ye - Ys(.5s+1)(.ls+1)
This open outer-loop driver/vehicle system has considerable low to mid frequency
lag. Applying the crossover model, it is apparent that a considerable amount of
low frequency driver lead in responding to the lateral position error would be
needed or else the driver would use a relatively low outer loop gain (K y) in order
to achieve an outer loop closure in accordance with the model. Data from simulator
and actual vehicle experiments show that drivers do the latter, i.e. use a relatively
low outer loop gain and rely on the inner loop function to provide the needed
lead equalisation or anticipation. Therefore, the outer loop describing function is
simply a gain, that is
(7)

An example value of K y for the assumed vehicle is about 0.2 heading com-
mand/ft of path error (1 rad/300 ft), or an outer loop crossover frequency KyVo
of about 0.27 rad/s. If the vehicle dynamic properties are other than these typical
passenger car examples, driver lead or lag equalisation can be used in the inner
loop to obtain the desired characteristics according to the crossover model.

17.3.1 Equivalent Single-Loop System for Steering Control

For measurement and data analysis purposes, since a useful form of the driver
model for directional control with steering is known, it is possible to identify
the parameters in the multiple-loop model in Fig. 17.3, using the single yaw dis-
turbance. The system in Fig. 17.3 can be expressed as an equivalent single-loop
system by using block diagram algebra on Fig. 17.3, resulting in the expression
rT
(8)
rd - 1 + Y1/tY;
s2
(yY U + s)'0

where the terms are defined above and in Fig. 17.3. This expression was originally
published in (McRuer et al., 1975).
 17. Review of Control Theory Models for Directional and Speed Control 299

FIGURE 17.4. Equivalent single-loop system.

An equivalent form of Eq. (8) can be written as

rr 1
(9)
rd 1+ G
which corresponds to Fig. 17.4, and where

YljrYI ( )
G == - 2 - YyUo +s (10)
s
and for s or W small

(11)

while for to large

y r
G- ljrY8 - Y yljr (12)
- s - ljr 8 .
The resulting combined driver/vehicle open-loop system, Eq. (10) is shown in the
Bode plot of Fig. 17.5. The following are interesting things to note in Fig. 17.5:

The outer loop driver gain (K y) defines a low frequency break point in the
amplitude ratio. This also results in an increasing phase lag at low frequency.
The crossover frequency (we) defines the closed-loop 0 dB line, and this becomes
the closed-loop bandwidth of the heading control loop. The crossover frequency
is equivalent to KljrK8.
The time delay term (e- r s ) results in a progressively increasing phase lag.
The vehicle yaw time constant (Tr ) appears as an additional high frequency lag
term.

These features of this example model application are shown subsequently in the
data for actual drivers.

17.4 Model for Speed Control with Accelerator Pedal

The driver model for speed control is similar to the model for directional control,
above, but simpler. The model of interest is for speed regulation, where the driver
uses the accelerator pedal to maintain a specified constant speed in the presence of a
speed disturbance. This model assumes that the driver perceives and operates only
on forward velocity. A possible multiple loop alternative could include an inner
loop of longitudinal acceleration, for situations where that cue can be usefully
300 Weir and Chao

Driver:
K =0.25
K;= 11300

40

30

20 20 dB/dec
~
'U
C 10
xII
'ii
(9
0

-10

-20 1
10. 10

<. KlJIK&

-90

-180
a
-8
rn
tn
if -270

-360 '---_'---...a...-'--'-....I.....lo-.........._--'---'---'-....&.--I.~L....l...___
_ _ L . _ _ _ ' _ _ . . . . L _ _ I........................

10.
' Freq(radlsec)

FIGURE 17.5. Open-loop combined driver/vehicle system bode plot for steering control.

perceived. The additional model for speed control of braking is beyond the scope
of this paper.
The speed control model is shown in Fig. 17.6. It has only a single loop with one
driver describing function element (Y u ) . The speed disturbance (Ud ) helps define
the driving task, and can be used to experimentally identify describing function
parameters. The dynamic model of the vehicle (YOU) is assumed to be known. The
driver describing function can be estimated as discussed below.
The first step in applying the crossover model is to examine the controlled
element (vehicle) transfer function. The transfer function relating forward speed
 17. Review of Control Theory Models for Directional and Speed Control 301

Speed Disturbance
Ud
r-------------------------
DRIVER :r------------------i----
VEHICLE
+
[IJU
I

8r : 10\ U
Desired r-r-r-r--:....... ~ 8 \(Y-"'-----. Actual
Speed :
I
+ Speed
I
I
I
I
I

~------------------------

FIGURE 17.6. Model for speed control with accelerator pedal.

to accelerator pedal position can be approximated by

yu _ Kf (13)
8 - (Tus + 1)'
where T; is the time constant associated with a change in vehicle speed.
This linear representation assumes (for simplicity) that positive and negative
changes in speed result from corresponding equivalent changes in pedal position.
Representative values in Eq. (13) for a passenger car at 50 mph are
K; = 1 ft/s per % pedal travel (14)
't; = lOs. (15)
So, Eq. (13) for this example is
u 1
(16)
Y8 = (lOs + 1)
Applying the crossover model and related manual control principles, a correspond-
ing form of the driver model is

1
Yu = x, (~ + TL ) e
-is

(TLs + 1) -is
= K; e. (17)
s
This involves proportional plus integral (trim) response, which reflects the lack of
very low frequency gain in the open-loop vehicle characteristic in Eq. (16) (it is not
K/s-like at very low frequency). A representative set of driver model parameters
for this speed control task is
K; = 0.3 % pedal travel/ft/s
TL = 12 s
i = 1.7 s

The drivers lead TL approximately cancels the vehicle lag Tu . The result is a rela-
tively low bandwidth driver/vehicle system for speed control, with a relatively
 302 Weir and Chao

20 FIGURE 17"7" Open-loop

driver/vehicle system bode
10
plot for speed control.
:c
"C
0 -~~-OdB

C -10
"(ij
~
-20

-30

-40
10-2

C>
(1)
"C
-90
CD
en
ca
..c:
CL
-180

-270
10-2
Frequency (rad/sec)

large time delay and a crossover frequency of about 0.3 rad/s, as shown in
Fig. 17.7.
Interestingly, the earlier literature and data involving the crossover model do
not address the low frequency integration form of Eq. (17). Therefore, the values
shown above are taken from more recent driving simulator studies involving this
particular control task, the large time delay in particular. More data examples are
given subsequently.

17.5 Experimental Data

Driver behaviour studies using both the DRI Driving Simulator and an actual test
vehicle provide example driver describing function data for a variety of driving
tasks.

17.5.1 Driving Simulator Measurements

The DRI Driving Simulator has been used extensively in studies of driver behaviour
and driver/vehicle interaction. Operational since 1993, it is a large-scale research
grade simulator, which features an automobile cab with instrumented controls
and displays, an electromechanical steering loader, computer-generated graphics
roadway scenes with a 180 field of view using three projectors as well as visual
0

scene graphics and animation accomplished on an SG ONYX IR using proprietary

 17. Review of ControlTheory Models for Directional and SpeedControl 303

FIGURE 17.8. DRI Driving Simulator.

DRI software . Motion cues are provided by a large hexapod motion system, with
a secondary 4DF seat motion system to provide road vibration and other ride cues
(see Fig. 17.8).
A driving simulator is useful for driver behaviour studies because it is easy to
study a range of conditions, the conditions are well defined and repeatable, it is
easy to change tasks and parameters, experimental measures are easy to obtain, and
a simulator is efficient and has low risk compared to studies using actual vehicles.

17.5.1.1 Steering Control

Driving simulator study results are available for the 10 typical but more mature
driver subjects listed in Table 17.1. The task was to maintain the vehicle in the
centre of the lane on a straight road in the presence of a random appearing lateral
disturbance. Similar to crosswind gustiness, the disturbance consisted of a sum
of non-harmonically related sinewaves to facilitate the describing function mea-
surements. The drivers were also controlling speed in the presence of a random
appearing longitudinal disturbance. Each run lasted about 2 min, including startup,
and 60 s of steady-state driver/vehicle response data from the middle of the run
were analysed. The subjects were doing only this primary driving task.
The resulting describing function parameters for the 10 subjects are shown
in Table 17.2. Two runs are shown for each subject. The average values over
304 Weir and Chao

TABLE 17.1. Driver subjects in an example steering

control study.
Subject Occupation Gender Age
1 Engineer M 46
2 Ship captain F 51
3 Small businessman M 55
4 Publisher M 44
5 Homemaker F 48
6 Administrator M 47
7 Retired police officer M 57
8 Postal supervisor F 54
9 Insurance agent M 55
10 Insurance agent F 54

the 20 runs are in the bottom row. The data were analysed using the equivalent
single-loop method described in Fig. 17.5. The first data column is the crossover
frequency (we). The second column (WI80) is the frequency where the open-loop
driver/vehicle phase angle is -180. The next column is the phase margin (<Pm),
which is a measure of closed-loop stability and damping. Combining We and
<Pm gives an estimate of the driver time delay (r) and this is shown in the next
column. Note that the driver time delay shown is the result of subtracting the
0.1 s yaw time constant, and an additional 0.07 s simulator graphics delay,
from the total measured driver/vehicle system delay. The final column is the
linear coherence (r02), which is the ratio of the driver's control action which is
linearly correlated with the disturbance input to the driver's total steering control
activity. The parameter values in the table illustrate both the within-subject and
between-subject variability for typical driver subjects.
The previous literature on the crossover model, and its occasional application
to driving, typically show values that differ somewhat from those in Table 17.2. In
particular, the example data for mature (age 44 to 57) typical drivers in Table 17.2
show somewhat lower crossover frequencies (gains) and larger time delays. In ad-
dition to possible age effects, these differences may reflect the effect of such things
as multiple tasking (speed regulation at the same time), use of a steering wheel
(increasing the arm-hand closed-loop neuromuscular delay), and lower levels of
skill (compared to pilot subjects, for example). Note that the main difference is a
difference in the time delay (r). To the extent that the subjects use the same phase
margin, the crossover frequency is a consequence of the time delay, as expressed in
the crossover model. Note also that Subject 1, who is an engineer and a pilot, had
larger We and lower t . These contemporary data and others like it, reflecting more
mature drivers and typical driving tasks, may be pertinent to future applications of
the driver models in addition to the values used in the past.

17.5.1.2 Speed Control

Limited describing function measures for speed control area also available, and
example data for two subjects are given in Table 17.3. The characteristics for
 17. Reviewof Control Theory Models for Directionaland Speed Control 305

TABLE 17.2. Example describingfunction parameters for steering

control.
We w180 cPm t
Subjects (rad/s) (rad/s) (deg) (s) r02

1 1.6 3.6 29 0.51 0.57

1 1.4 3.4 41 0.44 0.52
2 1.2 3.1 50 0.42 0.53
2 1.0 2.7 29 0.95 0.48
3 0.9 4.5 43 0.79 0.20
3 0.7 3.8 29 1.36 0.65
4 1.1 2.9 34 0.72 0.66
4 0.9 3.3 35 0.95 0.76
5 1.2 3.8 33 0.67 0.66
5 1.1 3.6 30 0.79 0.69
6 1.5 5.6 40 0.40 0.63
6 0.8 4.1 37 0.96 0.70
7 0.8 3.5 50 0.73 0.60
7 0.7 3.8 44 1.04 0.65
8 1.3 4.0 49 0.37 0.34
8 0.6 3.6 50 0.98 0.6
9 1.0 3.7 31 0.92 0.59
9 0.8 3.7 43 0.81 0.56
10 1.1 3.4 52 0.44 0.65
10 1.0 3.6 52 0.49 0.60
Average 1.0 3.7 40 0.74 0.58

Subject 1 are in Table 17.1. Subject 11 was a 33-year-old female business admin-
istrator. The task was similar to that described above for the steering control data.
A random appearing longitudinal disturbance consisting of a sum of sinewaves
was used to define the task and facilitate the describing function measures. The
describing function calculations were made over an interval of 48 s imbedded in a
2-min run.
The columns in Table 17.3 are similar to those of Table 17.2. Analysed in
terms of the crossover model, and compared to typical results for other tasks in the
manual control literature, the example results for speed control show relatively low
crossover frequencies, large time delays, and large stability margins, for reasons
previously noted. Note that data for Subject 1 appear in both Tables 17.2 and
17.3, which provides an interesting direct comparison of control activity in the
two simultaneous control tasks. Although more data would be useful, the data of

TABLE 17.3. Example describingfunction

parametersfor speed control.
We w180 cPm 'T
Subjects (rad/s) (rad/s) (deg) (s)

1 .35 1.0 65 1.7

11 .3 0.8 50 2.3
306 Weir and Chao

TABLE 17.4. Comparison of steering control describing function

measures in driving simulator and actual vehicle, Subject 1.
Run length We wI80 <Pm 'T
Method (s) (rad/s) (rad/s) (deg) (s)

Driving simulator 60 2.2 4.1 36 .43

Actual vehicle 40 2.8 5.9 53 .23

Table 17.3, representing typical drivers in this task, may provide guidance in future
applications of the model.

17.5.2 Actual Vehicle Measurements

Data are also available from an actual vehicle configured for the input of a yaw
disturbance, similar to the disturbance used in the driving simulator studies de-
scribed above. For comparison purposes, the same random appearing directional
disturbance signal was used.
Example steering control describing function data for Subject 1 are shown in
Table 17.4. The driving simulator values shown are the average of 7 runs, 60 s
each. The actual vehicle values are the average of 2 runs, 40 s each. These limited
and preliminary example data show an interesting comparison between driving
simulator and actual vehicle, in that the measured time delay is less in the latter
case. The vehicle dynamics (yaw time constants) were approximately the same in
the two cases. The measured simulator r can be reduced by the 0.07 s graphics lag,
as noted previously, which reduces 0.43 to about 0.36. The remaining difference of
about 0.1 s can probably be attributed to the fact that these particular simulator runs
were made fixed base (motion turned off), while the actual car runs had motion.
The observed difference of 0.1 s is consistent with previously reported effects of
motion versus no motion in simulator studies (e.g. Stapleford et al., 1969). Note,
also, that the describing function parameters for this subject in the simulator in
Table 17.4 show smaller r and larger crossover frequency, than did the average of
the typical subjects in Table 17.2. Again, this probably reflects a higher level of
skill on the part of Subject 1. The data of Table 17.4 also serve as an interesting
further confirmation of the DRI simulator's validity, as has been demonstrated
repeatedly in the past.

17.6 Example Directional Control Application

An example application of the steering control model in Fig. 17.3 can be given by
the analysis of a hypothetical high-speed driver/vehicle stability problem. Consider
an inappropriately overloaded large sports-utility-type vehicle with the following
hypothetical example values for cg location (wheelbase), mass, yaw moment of
inertia and front and rear tire cornering stiffness, respectively:
 17. Review of Control Theory Models for Directional and Speed Control 307

a = 3.5 ft (1.1 m)
b = 5.5 ft (1.7 m)
m = 185 slugs (2700 kg)
I = 4000 slug/ft" (5400 kgm/)
YI = 4000 lb/rad (18 kN/rad)
Y2 = 5000 lb/rad (22.4 kN/rad)
A simple set of 2 of freedom lateral-directional equations for the vehicle is given
by

s - Yv
(18)
[ -Nv

where v is the lateral velocity, r is the yaw rate, 8 is the steer angle, V 0 is the
forward velocity, s is the Laplace transfer variable and the coefficients are
-2
Yv =- (YI + Y2 )
u,
2
Yr =- (bY2 - aYt)
u;
2
N; =- (bY2 - aYI )
IVo

NT = -2 (a 2 Y1 + b2 y 2)
IVo
2
Y8 = -Y1
m
2a
N8 = -YI.
I
Assume a speed of 120 ft/s (130 kmlh). The result is a relatively poor handling
example vehicle, with the assumed excessive load and inadequate tires, but it can
serve to illustrate the analytical use of the driver model. Entering the numerical
parameters listed into the equations above gives the following vehicle transfer
function for heading control (from Eq. 4):
1/1 7 (s + 1.16)
(19)
Y8 = S [s2 + 2 (.3)(2.7) s + (2.7)2]'
The damping ratio is 0.3 and the undamped natural frequency is 2.7 rad/s in
the characteristic equation, represented by the complex conjugate pair. While the
Eq. (2) vehicle approximation for yaw rate was a simple high frequency lag, the
slightly more complete form of Eq. (19) replaces that with a lead-double lag
combination. The equivalent yaw time constant, Ti., and phase lag contribution for
frequencies less than 2.7, would now be about 0.22 s, instead of the 0.1 s example
used with Eq. (2). The driver/vehicle open-loop response properties (see Fig. 17.3)
for this example become

(20)
308 Weir and Chao

20 .------.--...---.--~.........____r___,___,~....,...,._r_-.....___T'""____.__._......_r_r_r_l
FIGURE 17.9. Example
driver/vehicle system bode
10
plot for heading control.
:c
"0
0
~ -10
CJ -20

-30
-40 '------'----'--'---'--'"~O.__"______'__~~O.__"______'__..............................

10-2

0)
Q)
-90
~
"0
Q) --------------------"""\\ r = O
CJ)
ct1
..c:
Q. -180

\
\ 't= 05
.
-270 L-.._"______'__.&......J.-,.i~O'__"______'__iL_...L......l ~
..............." ' _ _ _ _ _ _ ' _ _ _ _ ' _ _...............

10-2
Frequency (rad/sec)

where Y81/1 is given by Eq. (19). The corresponding frequency response plot for
Eq. (20) is shown in Fig. 17.9, in the manner of Fig. 17.2. A driver time delay
of 0.5 s is used in this example, and its effect is shown by the lower phase angle
curve.
The next step is to estimate the driver gain K 1/1. Fig. 17.9 shows that the driver gain
is limited by the lightly damped second order made at 2.7 rad/s. If the driver's gain
is increased so that the closed-loop OdB line intersects the peak in the amplitude
ratio plot at to == 2.7 rad/s, the closed-loop driver/vehicle system becomes unstable.
Therefore, a lower driver gain has to be used resulting in relatively low values of
closed-loop bandwidth and directional stability. This is illustrated, alternatively,
in the corresponding root locus plot of Fig. 17.10. The branch of the locus from
the quadratic pole moves into the right half plane and becomes neutrally stable
or unstable for a relatively low level of driver gain K 1/1. This value of gain would
cause the vehicle to move back and forth in the lane. Figs. 17.9 and 17.10 both
show that this oscillatory behaviour for this hypothetical vehicle configuration is
due to the driver's time delay. The resulting phase lag causes the closed-loop poles
to move towards a region of instability as the driver's gain is increased (as the
driver tries harder and makes larger control inputs). This type of system response
has sometimes been characterised as a 'driver induced oscillation'. Note that there
is little point in analysing the outer path loop, since adding that feedback would
 17. Review of Control Theory Models for Directional and Speed Control 309

FIGURE 17.10. Example root 6

locus plot for heading control.
5

4 r= 0

Cf)
x

3
1.. ,.,..-.-..-"
t > 0.5
0> 2
ctl
E

o
-1

-2
-6 -4 -2 o 2 4 6
RealAxis

have little effect on the vehicle's behaviour and lane-keeping performance in this
example.
A more detailed analysis would show that lag equalisation might improve the
closed-loop bandwidth in Fig. 17.9 a small amount. But, for the purposes of this
example, it assumed that the driver does not provide such smoothing. For this hypo-
thetical poor vehicle configuration, corrections of heading errors by the modelled
driver can only be done using small steering inputs. Aggressive or large amplitude
steer corrections would tend to result in oscillatory directional behaviour, as noted
above. In other words, the quasilinear model says that if the driver tries harder it
only worsens the stability, in this case. The driver could perhaps adapt a nonlinear
control behaviour, wherein the frequency of the oscillatory response of the vehicle
is recognised and the driver makes a properly timed and predicted discrete steering
correction to reduce the oscillation, in effect reducing or eliminating the effect of
his or her time delay. Another possibility is to hold the steering wheel approxi-
mately fixed and let the directional oscillations damp out if the vehicle alone has
positive path damping.
It should be emphasised that this hypothetical high speed vehicle example is
only to illustrate a possible use of the model. If the vehicle were to slow down
the damping would improve. In addition, or alternatively, the example vehicle
configuration could be readily changed to eliminate the undesirable effect shown,
by reducing the mass and inertia (the load), or by increasing the tire cornering
stiffness characteristics (more suitable, higher performance tires).
This example shows how driver/vehicle response properties can be predicted
analytically. This type of comparatively simple and robust model and analysis
can assist the understanding of a problem or issue. Such analysis also shows the
contribution of the several driver and vehicle parameters to the resulting response,
and it suggests how sensitivity analyses can be performed to improve or optimise
310 Weir and Chao

driver/vehicle response properties. For instance, one could analyse the effect of
varying the driver time delay in this example.

17.7 Discussion
Models for driver directional and speed control have been presented, and their
application has been illustrated. These are based on models in the classic literature
and manual control principles. Data have been presented for typical values of the
model parameters and to show how these parameters vary depending on the task,
the experimental venue and across driver subjects.
The quasilinear models described herein are best suited to regulation tasks where
the driver is compensating and minimising errors on straight and gradually curving
roads at an approximately constant speed, and these conditions represent a large
percentage of driving. Other models that are available for other driving tasks
such as rapid manoeuvres, obstacle avoidance, near limit turning and/or braking
manoeuvres, etc., are beyond the scope of this paper.
Driver models of the sort described herein have a wide variety of applications.
For example, they can be used to characterise the driver element in R&D-related
studies of vehicles, of devices-in-the-vehicle, or vehicles augmented by devices.
The models can be used to interpret driver behaviour or driver/vehicle response
and performance data. They can be used to represent the driver or driver/vehicle
system in simulations involving one vehicle or multiple vehicles. Perhaps most
usefully, they serve to quantify driver behaviour and driver/vehicle response and
performance in engineering terms, and in a way that both predicts what the driver
will do and quantifies data resulting from what the driver has done.

References
Allen, R.W., Rosenthal, TJ. and Hogue, lR. (1996). Modelling and Simulation of
Driver/Vehicle Interaction. SAE Paper 960177.
Brackstone, M. and McDonald, M. (1999). Car following: A historical overview. Trans-
portation Research Part F, 2, 181-196.
Brown, T.L., Lee, J.D. and McGehee, D.V.(2001). Human performance models and rear-end
collision avoidance algorithms. Human Factors, 43, 462-482.
Donges, E. (1978). A two-level model of driver steering behaviour. Human Factors, 20,
691-707.
Fancher, P.S. and Baraket, Z. (1998). Evolving model for studying driver-vehicle system
performance in longitudinal control of headway. Transportation Research Record No.
1631 (pp. 13-19).
Godthelp, J. (1985). Precognitive control: Open- and closed-loop steering in a lane change
maneuver. Ergonomics, 28,1419-1438.
Hoffman, E.R. (1975). Manual control of road vehicles. Vehicle System Dynamics, 5, 1-2.
Levison, W. and Cramer, N. (1995). Description of the integrated driver model. Report
FHWA-RD-94-092. FHWA, US DOT.
 17. Review of Control Theory Models for Directional and Speed Control 311

MacAdam, C.C. (2003). Understanding and modelling the human driver. Vehicle System
Dynamics, 40, 101-134.
McRuer, D.T. and Weir, D.H. (1969). Theory of manual vehicular control. IEEE Transac-
tions, MMS-I0(4). (Also Ergonomics, 12 (5), 1969.)
McRuer, D.T. and Krendel, E.S. (1974). Mathematical models of human pilot behaviour.
AGARDograph No. 188. NATO Advisory Group for Aerospace Research and Develop-
ment.
McRuer, D.T., Weir, D.H., lex, H.R., Magdaleno, R.E. and Allen, R.W. (1975). Measure-
ment of driver/vehicle multiloop response properties with a single disturbance input.
IEEE Transactions, SMC-5(5).
McRuer, D.T., Allen, R.W., Weir, D.H. and Klein, R.H. (1977) . New results in driver steering
control models. Human Factors, 19(4),381-397.
Sheridan, T.B. and Ferrell, W.R. (1981). Man-Machine Systems, Information, Control, and
Decision Models ofHuman Performance. MIT Press, Cambridge, MA.
Stapleford, R.L., Peters, R.A. and Alex, F.R. (1969). Experiments and a model for pilot
dynamics with visual and motion inputs. NASA CR-1325.
Weir, D.H. and McRuer, D.T. (1968). A theory for driver steering control of motor vehicles.
Highway Research Record, 247, 7-39.
Weir, D.H. and McRuer, D.T. (1973). Measurement and interpretation of driver/vehicle
system dynamic response. Human Factors, 15(4),367-378.
Weir, D.H. (1973). Driver/vehicle response and performance in the presence of aerodynamic
disturbances from large commercial vehicles. In First International Conference on Driver
Behaviour, Zurich, Switzerland.
Weir, D.H. and DiMarco, R.I. (1978). Correlation and evaluation of driver/vehicle direc-
tional handling data. SAE Paper 780010.
Weir, D.H., Klein, R.H. and Zellner, J.W. (1982). Crosswind response and stability of car
plus utility trailer combinations. SAE Paper 820137.
 VII
Simulation of Driver Behaviour
18
Cognitive Modelling and
Computational Simulation of
Drivers Mental Activities
THIERRY BELLET, BEATRICE BAILLY, PIERRE MAYENOBE AND OLIVIER
GEORGEON

18.1 Introduction: A Brief Historical Overview on Driver

Modelling

Several car driver models are available in the literature. From an historical point
of view, it is possible to identify three main phases, for the last 40 years, in this
research area. The works of Me Knight and Adams (1970), centred on task anal-
ysis, are typically representative of the studies carried out during the 1970s. The
authors proposed a taxonomy of the main driving tasks (e.g. accelerating, steering,
overtaking, lane changing) organised in nine categories (e.g. basic control tasks,
tasks related to traffic condition). This work closely parallels the research of Allen
et al. (1970), who divided the driving task in three levels: the microperformance,
the situational performance and the macroperformance. These levels differ both
according to their time scale and with regard to the kind of cognitive activity re-
quired. At the microperformance level, most of the actions are automated skills.
Steering and speed control are the main subtasks. Feedback loops, concerning
driving action implemented at this level, are very short (on the order of seconds).
The macroperformance concerns the trip planning and the route finding during the
trip. It corresponds to slow conscious processes requiring cognitive resources. The
time scale can be hours at this level. Between these two levels, situational perfor-
mance corresponds to the analysis of the road environment and to the selection of
relevant behaviour in the current situation and traffic conditions. Performance at
this level is determined by the driver's perception and understanding of the driving
context.
These types of studies are descriptive in nature and aim to propose an exhaustive
taxonomy of the different kinds of driving subtasks. This approach provides an
interesting framework to analyse driving behaviours, but is poor in terms of men-
tal activities study. It provides models of the driving task more than model of the
drivers. During the 1980s, the main paradigm was the human information process-
ing approach. Several models were developed in order to explain accidents. Most
of them are focused on the notion of risk: the risk threshold model of Naatanen
and Summala (1976), Wilde's risk homeostatis model (1982) and Fuller's threat

315
316 Bellet et al.

avoidance model (1984). These models describe drivers activity as a regulation

task in order to maintain a subjective (i.e. estimated) level of risk underneath of a
given target level (i.e. accepted risk). Consequently, they more particularly insist
on the driver's motivations and risk assessment. One of the most interesting model
developed during this decade is the hierarchical risk model of Van der Molen
and Botticher (1988). This model aims to provide 'a structural framework per-
mitting the description ofperception, judgement and decision processes at every
level of the task'. Based on Michon's works (1985), the model is hierarchically
structured with a strategic, a tactical and an operational level. This three-level
hierarchy parallels the skill-rule-knowledge (SRK) model of human information
processing proposed by Rasmussen (1983). It also evokes the division of driving
tasks proposed by Allen et al. (above-mentioned). Nevertheless, by contrast with
King approach, the hierarchical risk model is not focused on the driving task it-
self, but on the drivers' mental activities (as SRK). The model describes several
cognitive mechanisms such as the formulation of a risk judgement, expectations
or decision making. However, each of these processes is only modelling via really
simplified equations. Moreover, information processing is strongly sequential in
this model, and it seems not appropriate to describe the functioning of the human
cognitive system. Lastly, while insisting on the role of internal representations in
decision-making processes, the authors do not describe these mental structures,
and the cognitive processes involved in these mental models construction are also
not defined. Therefore, we will see that COSMODRIVE (cognitive simulation
model of the driver) has been more particularly developed in order to modelling
these cognitive mechanisms.
During the 1990s, a third generation of models was introduced with the aim to
provide computer simulation of the driver. It is possible to identify at least two
types of simulation models: models of performance (Levison, 1993) and cognitive
simulation models (Wierda and Aasman, 1992; Salvucci et al., 2001; Krajzewicz
et al., 2004). Except that they both propose computational models, these two ap-
proaches are different in their theoretical foundations as well as in their application
areas. The firsts one are essentially focused on human behaviour, with the aim to
simulate it and predict performances (i.e. predictive models) and not the internal
processes implemented behind behaviour. By contrast, the main objective of the
cognitive simulation models (or explicative models) is to describe and simulate
internal states and cognitive processes carried out during driving. Some of these
models can be also used as predictive models of human performance, even though
it is not their main goal. This distinction between predictive versus explicative
models is very important in terms of epistemology (Thom, 1991). It determines
the computational modelling phase: for a performance model, what is the most
important is the final result, irrespective of how the processes are implemented.
These models are generally limited to one (or more) black box(es) whose inputs
correspond to information coming from the road environment, while the outputs
correspond to performances. The implementation choices, at the level of the 'black
box', are essentially determined by technical criteria (e.g. calculation time), with-
out considering the real nature of the processes involved in the human cognitive
 18. Cognitive Modelling and Computational Simulation 317

system. Only the matching of the model's prediction (predicted behaviour) and
the facts observed in the real world (effective behaviour) is considered. This kind
of model is more appropriate in the framework of a behavioural approach. On the
other hand, cognitive simulation models are not just focused on the probability
of occurrence between a situational context (the characteristics of a traffic situa-
tion) and driving behaviour. The way the processes are implemented leads to an
homomorphism with the human cognitive processes. This approach is heavier in
terms of human modelling, but allows a better understanding of the cause-effect
relationships between behaviour and the driving conditions.
The purpose of this paper is not to judge these two approaches in terms of
correctness. Both of them are relevant, but can differ in their goals. Performance
models are more appropriate if the aim is only to predict behaviour. In our case,
the goal is to evaluate theoretical approach and experimental results concerning
cognitive processing in the driving context, with the aim to provide a deep under-
standing of driver's cognition. From this point of view, the explicative modelling
approach was needed.

18.2 COSMODRIVE Model

COSMODRIVE is based on several empirical data coming from drivers' activities
analysis in real driving situations (Bellet, 1998) or from lab experiment results
(Bailly et aI., 2003). Empirical data are focused on the activity of turning left
at an urban road junction. Two series of observations were carried out: on-site
observations (outside the car) and on-board observations (on an experimental car
equipped with sensors a video cameras), followed by interviews with the drivers
(based on the video film recorded during the trip). Lab experiments are based on
video film and the main results obtained are described in Section 18.4.

18.2.1 Cognitive Architecture of COSMODRIVE

From a fundamental point of view, COSMODRIVE's cognitive architecture dis-
tinguishes two main types of information processing and storage structures (Fig.
18.1): the long-term memory (LTM), modelling as knowledge bases that hold
driver's knowledge, and the working memory (WM), modelling as blackboards
that contain mental models of the driving situation. Several cognitive processes
(or agents) are in charge (a) to activate (retrieval) or store (acquisition) knowledge
in the LTM, (b) to process information coming from the road environment and
integrate it in the WM, (c) to carry out reasoning (e.g. decision, anticipation, ac-
tion planning) and (d) to manage cognitive resources. Mental representations build
in the WM playa central role in the driver's cognitive activity. They correspond
to the driver's understanding of the driving situation and provide mental models
that are more or less true to reality. It is on the basis of this 'representation of the
world' that the driver takes decisions when driving. COSMODRIVE endeavours
to simulate all the processes involved in generating and using these mental models.
318 Bellet et al.

HUMAN COGNITVE SYSTEM

FIGURE 18.1. Simplified view of the cognitive architecture of COSMODRIVE.

In terms of level of control, COSMODRIVE's architecture is based on Mi-

chon (1985) and Van der Molen and Botticher (1988) approaches. The model
distinguishes a strategic module, a tactical module, an operational module and an
emergency management module (the latter two modules being part of the risk hi-
erarchy model). Three other modules complete this basic framework: perception,
execution and control and management modules.
The strategic module includes trip planning and route finding activities and in-
tegrates travel constraints (e.g. to reach the destination at a given time) and global
goals (e.g. to find a service station). The initial plan can change during the trip, if
new goals are integrated (e.g. to find a restaurant, if I am hungry). The results of rea-
soning carried out at the strategic level are given to the tactical module in the form of
local goals to be reached at varying times. The tactical module can be considered as
a generator of mental models of the current driving situation. Driving actions (e.g.
overtaking a car, crossing an intersection) are also planned and decided at this level.
This module includes several cognitive processes such as decision making, driving
context categorisation, place recognition or anticipation. The tactical module also
has an anticipation process that generates anticipated representations (ARs) from
the current state of the world. Each AR corresponds to a possible evolution of the
current situation. These anticipations playa decisive role in choosing the driving
actions to be implemented. Once they have been selected, the actions are transmit-
ted to the operational module. The operational module is built as a multi-agent
system with a set of autonomous operational units, specialised in implementing
elementary driving subtasks (steering control, speed control and distance main-
taining between vehicles). From a cognitive point of view, mental computation are
generally sub-symbolic at this level, by contrast to the strategic and the tactical
reasoning based on symbolic computation. In any case, operational units cooper-
ate in order to define the precise action plan of driving decision determined at the
tactical level. Then, driving behaviour will be carried out by the execution module,
via vehicle controls (pedals, steering wheel, etc.), in order to progress on the road
infrastructure. Another function of the operational module is to assess the risk of
 18. Cognitive Modelling and Computational Simulation 319

accident. Specific reasoning is in charge to identify a danger and, if necessary,

to activate (sending a message) the emergency management module. In this case,
tactical and strategic modules are 'disactivated', until an emergency reaction has
been selected in memory and the problem solved. Once the driving situation is
under control, the emergency module sends a message to the control and manage-
ment module in order to reactivate the tactical and strategic modules. Then, the
emergency management becomes inactive until the occurrence of a new incident.
COSMODRIVE also integrates two other modules. The perception module is the
interface between the road environment and the model. It simulates information
processing of sensorial data before integration in the tactical module (i.e. in the
mental model of the driving situation). Two modes of information processing are
implemented. The first one is 'data-driven' (bottom-up procedure based on a set
of filters) and allows cognitive integration of unexpected information. The sec-
ond one is 'knowledge-driven' (top-down mechanism) and is implemented via a
perceptive exploration agent in order to examine the road environment according
to the perceptive expectations included in the tactical representation or requested
by operational level agents. From a computational point of view, perception is in
charge to manage the perceptive queries (request for information to be obtained)
coming from the different cognitive processes active at a given time. At this level,
the aim is to determine the order of priority of these queries and, on this basis, to
specify the perceptive exploration strategies for exploring the road scene. Informa-
tion collected is then transmitted to the querying cognitive processes. A detailed
description of this module is available in Mayenobe (2004). Lastly, the control
and management module is in charge to simulate the management mechanisms of
cognitive resources. Like the human driver, COSMODRIVE is limited in his atten-
tional capacities. The main task of control and management is to share cognitive
resources between the different cognitive processes implemented at a given time.
Schematically, the resource allocation procedure starts with the demand, from
each active process, for a quantity of resources necessary to perform its current
task. This 'demand' occurs by sending a query to control and management. This
module analyses all the demands and the allocation is done, in return, via the pa-
rameterisation of the demanding process. Control and management also manages
the transition between the 'abnormal situation'/'normal situation' modes: as soon
as a danger is detected, all cognitive resources are allocated to the emergency mod-
ule, and the tactical and the strategic modules must wait the solving of the critical
situation to have cognitive resources. The central module of COMODRIVE in the
field of this chapter is the tactical module.

18.2.2 The Tactical Module

Driver's behaviour is not directly based on the objective state of the world, but
on a mental model (internal representation) of the driving situation. This mental
model is built in a working memory from perceptive information extracted from
the road scene and from permanent knowledge stored in the long-term memory.
At the tactical level, this mental model provides a meaningful and self-oriented
320 Bellet et al.

understanding of the reality, including anticipations of potential evolutions in the

current driving situation. From this point of view, it corresponds to the driver's situ-
ation awareness, according to Endsley's definition of this concept: 'the perception
of the elements in the environment within a volume of time and space, the com-
prehension of their meaning, and the projection of their status in the near future'
(Endsley, 1995). Moreover, this mental representation is 'action-oriented' (i.e. the
driver does not passively observe the road scene as a 'spectator', but as an 'actor').
It constitutes an operative image (i.e. a functionally deformed view of the reality;
cf. Ochanine, 1977), a 'goal driven' model of the road environment. Once built,
these mental models generate perceptive expectations, guide the road environment
exploration and the new information processing, orientate decision making and
lastly, determine all driving behaviours carried out by the driver. From this point
of view, tactical representations are a key element of the driver's cognition, and an
erroneous representation means, potentially, decision-making errors and unsafe
driving actions. The central objective of COSMODRIVE is the computational
simulation of the cognitive processes involved in this mental model building.

18.2.2.1 Driving Frames: A Framework for Modelling Mental Models

In order to study mental model, drivers' activity observations have been done dur-
ing a specific task: 'turning left at a urban crossroads with traffic lights' (Bellet,
1998). Several data have been collected and analysed (driver's visual strategies, ac-
tions performed, vehicle position and speed). All these data have been considered
taking into account the traffic conditions (other vehicles positions and manoeu-
vres). Moreover, driver interviews were conducted after this driving task (based
on the video film collected during the trip) with the aim to collect drivers' opinion
concerning their behaviours, their strategies, the perceptive information used, and
the knowledge involved for decision making. As a result, it appears that the drivers
use afunctional representation of the road environment to drive the car and to adapt
their behaviour according to the traffic conditions. A specific formalism - based on
the frame theory of Minsky - has been defined at LESCOT for these mental repre-
sentations modelling (Bellet et aI., 1999). Frame refers to psychological concepts
such as 'schema' (Bartlett) or 'scheme' (Piaget). In a classic paper, Minsky (1975)
defines frames as complex data-structures for representing situation knowledge
stored in human memory. The main advantage of the frame formalism is the pos-
sibility to integrate, in the same structure, declarative and procedural knowledge.
Consequently, frames can describe objects, situations, events, sequences of events,
actions and sequences of actions. Moreover, frames act like models of the world
that guide the interpretation of sensorial inputs and the perceptive explorations of
the environment. Frame approach has been used in different type of human opera-
tors model, more particularly in the industrial control process area (e.g. Cacciabue
et al., 1992).
Driver's mental model has been described in COSMODRIVE as driving frames.
A driving frame is a mental representation of the road scene, functionally structured
according to the goal pursued in this particular infrastructure. Figure 18.2 presents
 18. Cognitive Modelling and Computational Simulation 321

r" '-;;x6 :::'1 - - - - high sp eed v ehicle?

t it
l j"fex2tl
v ehi cle? ~ - 1~ III ~ ' D
\- -"' [ 'P i ' ~

'FD'I \f ~ll l l l l l mll l

I

Ii {J
ex? : rr-rr--:-:
ped estr ianf ,--~i I
:--; ex
ZJb2
=
=
c=J ~
=
=
= LI
--,
-~
- - -

-'

=
=
=
= -
:
ex21 color of the
1l traffic lights ?
L. _ ._jJ
c::J Initial State ex.!i : ped estri an ?
~ Local States
_ Final State

c:::J Driving Zones (Zi)

C::J Perceptive Exploration
Zones (Ext)
Driving Path alternatives

FIGURE 18.2. Driving frame of an urban crossroadwith traffic lights.

the generic driving frame (i.e. driving knowledge) of an urban crossroad with the
goal 'turn on the left' (this frame must be instantiated with event occurrences to
provide a mental model of a particular driving situation). A frame is defined by a
global initial state, a global final state, a set of perceptive exploration zones and
a driving path . A state corresponds to the vehicle position and speed at a given
time . The global initial state corresponds to the initial position and speed of the
car when the driver approaches the intersection. The global final state corresponds
to the final position of the driver when he or she leaves the junction. Reaching
this last position constitutes the global goal of the driver in this crossroad. The
driving path is composed of a sequence of driving zones. It allows the driver to
pass from the initial to the final state . Each driving zone is described by three
elements: a local initial state, and a set of localfinal states and a set of actions. An
action allows the driver to pass from the local initial state to one of the local final
states. In concrete terms , if we consider the Fig. 18.2, the road infrastructure is
described as a set of two kinds of zones : the driving zones (Zi) and the perceptive
exploration zones (Exi). The sequences of driving zones constitute the driving
paths. In this particular case, two driving paths are possible: Z I, Z2, Z3a, Z4 and
Z I, Z2, Z3b 1, Z3b2, Z4. The perceptive exploration zones correspond to the part
322 Belletet al.

FIG URE 18.3. Three-dimensional modelling of driving frames in COSMODRIVE.

of the environment that the driver observes in order to detect events. An event
corresponds to an object occurrence in a specific zone (e.g. a vehicle in ex3) or
to identify a particular characteristic of this object (e.g. the colour of the traffic
lights in ex l ), In order to tum left in this infrastructure, the driver realises several
consecutive steps: at the beginning of ZI (global initial state), he observes the
colour of the traffic lights . If the traffic lights are green , he then proceeds to Z2
with the same speed . If not, he stops the car at the end of Zl. In Z2, the driver
checks for vehicle presence in the opposite lanes near the crossroads (ex3, ex4 or
ex5 zones), and the presence of any high- speed vehicle in ex6 zone . If there is no
vehicle, the driver goes to Z3a. Else, he stops the car at the end of Z3b 1, and waits
until there is no vehicle in the ex4 and ex5 zones . In Z3a or Z3b2, the driver looks
for any pedestrian in the ex7 zone. If a pedestrian is present, he stops at the end of
Z3a or Z3b2 while the pedestrian crosses the street. And then, he continues his path
in Z4, towards the global final state of the frame. From a computational point of
view (Bellet and Tattegrain-Veste, 1999,2003), driving frames has been modelled
in COSMODRIVE with the object-oriented modelling technique of Rumbaugh
et al. (1991) and have been implemented in Small Talk language. In a more recent
work (Mayenobe et aI., 2002, 2003 ; Mayenobe, 2004) , a three-dimensional (3-D)
version of driving frames has been also developed, allowing dynamic simulation
of drivers' mental models (Fig. 18.3).
In terms of human cognition, these 3-D models correspond to visuo-spatial rep-
resentations of the road scene , mentally used by the driver for decision taking and
action planning at the tactical level. Content of these mental models depends on
several things, like the current driver's goal, driving experience, attention sharing
problem, age and so on. Experimental results will be presented in the next part
of this chapter to illustrate the potential impact of a secondary task or driving
 18. Cognitive Modelling and Computational Simulation 323

FIGURE18.4. The 'relative zones' and COSMODRIVE simulation of vehicles' interaction

management.

experience effect on driver's situation awareness. In any case, these 3-D models
provide simulation of the driver's understanding of the road environment, including
infrastructure and events perceived from the pilot point of view. They also integrate
relative zones around the driver 's own car (Fig. 18.4). These relative zones corre-
spond to the part of the space that drivers must to keep around their car to avoid
collision or to comfortably interact with other road user. From a theoretical point
of view, the relative zone approach is partly based on Hall's concept of proximity
bubbles (Hall, 1966), which concerns social interactions between human s. In driv-
ing context, a same idea has been also proposed with the concept of safety margin
(Gibson and Crooks , 1938). Nevertheless, COSMODRIVE three-relative-zones
model is more particularly based on Kontaratos' work (Kontaratos, 1974) and
distinguishes the danger, the threat and the safety zones . These zones are called
'relative' because they are dependent on the car speed and position. They corre-
spond to the portion of space to be occupied by the future trajectory of the vehicle.
The lengths of these "relative zones" depend of the car speed, and correspond to
the distance covered by the vehicle during this period of time. According to Otha's
results (1993), the reserved values of the coefficients are, respectively, 0.6, 1.1 and
1.7 s for the zones of danger, threat and safety. However, inter-vehicular distance
regulation may vary according to the context, the driving strategies and the traffic
density .
Moreover, 3-D mental models also integrate a set ofremarkable points (i.e.
restricted set of points corresponding to specific landmarks, like comers of the
roads or intersection centre; cf. see Fig. 18.3) needed to geometrically match the
mental model and the road infrastructure (i.e. objective reality as perceived by
the driver). Once a driving frame is matched with reality (Fig. 18.5), the driver
324 Belletet al.

FIGURE 18.5. From reality (left) to mental modelsimulatedwith COSMODRIVE (right).

can mentally use the driving path of the frame to mentally simulate trajectories
of vehicles , according to its own trajectory. This mental simulation mechanism
provides anticipations, which correspond to drivers ' expectations.
Lastly, at a lower level (which corresponds to the operational and the execution
modules of COSMODRIVE) driving frames are implemented through the method
of pure-pursuit point developed for automatic car driving (Amidi, 1990) and used
by Sukthankar (1997) for drivers' situation awareness simulation (Fig. 18.6). A
pure-pursuit point is defined to be the intersection of the desired vehicle trajectory
and a circle of radius (I), centred at the vehicle's rear axle midpoint (assuming front
wheel steer) . Intuitively, this point describes the steering curvature that would bring
the vehicle to the desired lateral offset after travelling a distance of approximately 1.
Thus the position of the pure-pursuit point maps directly onto a recommended
steering curvature: k = -2xll , where k is the curvature (reciprocal of steering
radius), x is the relative lateral offset to the pure-pursuit point in vehicle coordinates
and I is a parameter known as the look-ahead distance.

18.2.2 .2 Architecture of the Tactical Module

The whole activity of the tactical module of COSMODRIVE is dedicated to the
elaboration and the handling of these driving frames , as mental models of the cur-
rent driving situation. Computationally, the tactical module is implemented as a
multi-agent system based on a blackboard architecture: different processes or cog-
nitive agents process the information in parallel and exchange data (read/write) via

!
-_._---_._-;-_.._.__.__._-
iI relative lateral
offset (x)
_.._r
f pure-pursuit point(P) i
f i
; desired trajectory
i
!
Isteeringradius(r)

f
i

FIGURE 18.6. The pure-pursuitmethod.

 18. Cognitive Modelling and Computational Simulation 325

~~:~ ~;~'~~~]]]M#.i~~:: i:l:i;: ~~~*ffi:i~fmf~::::i,i':

: ::::'
Goals

FIGURE 18.7. General architecture of the tactical module.

common storage structures: the blackboards. Consequently, problem solving (e.g.

what driving action to take?) is distributed, resulting from the cooperation between
several agents: each cognitive agent has only a part of the required experti se to
find the solution. As the solving procedure progre sses, the product of each agent
reasoning is written in the blackboard . It then becomes available to the other agents
sharing this blackboard, and they can then use these results to furthering their own
reasoning. They can also modify the co ntent of the blackb oard and thus contribute
to the collective solving of the problem. Communi cation between the cog nitive
agents of the tactical module can also be done via message sending. Lastly, the
compl exity and speed of the processes performed are dependent on the cog nitive
resource s allocated to the solving agent by the control and management module .
The quantity of resources allocated can change according to the number of active
proce sses and their respective needs. A lower quant ity of resources can slow down
the agent computation, change the solving strategy or momentarily interrupt the
process underway. Synthetically (Fig. 18.7), the tactical module is compo sed of
two blackboa rds, three knowledge bases and five cognitive agents.

18.2.2.3 The Blackbo ards of the Tactical Module

The tactical module has two blackboards: the current state blackboard (CSB) and
the anticipation blackboard (AB). The CSB contains all information related to the
current driving situation. It include s a goals list, af acts list and a current tactical
326 Bellet et al.

representation (CTR). The goals list contains the 'local' goals to be attained in the
current infrastructure (e.g. turn left) and 'global' (or latent) goals to be reached
in the more or less long term (e.g. save time, because I am late). These goals
essentially come from the strategic module. The facts list contains all information
extracted from the road scene. This data comes from the perception module. The
CTR corresponds to the driver's mental model of the driving situation. It is built
from a generic driving frame activated in LTM, instantiated with data available
in the facts list and in the goals list. In contrast to the facts list, which only lists
events individually, the CTR specifies the spatial and dynamic relations existing
between these objects in the road environment. It provides a visuo-spatial mental
model of the road scene. Based on operative knowledge (i.e. driving frame), CTR
provides a functional, deformed, simplified and finalised representation of the
reality. On the one hand, only relevant pieces of information for the driver at
moment t, according to the pursued goal, are considered. On the other hand, some
data contained in the CTR could be not directly available in the environment at
moment t (e.g. data produced by inference, memorised previously, default values
of the driving frame). Once generated, the CTR provides a plan of the driving
activity. It orients new perceptive information collection and processing as well as
the decision making.
The AB holds ARs derived from the CTR by the anticipation agent. Each AR
constitutes a possible evolution of the current driving situation. ARs are organised
in a tree, whose root corresponds to the CTR, nodes correspond to the potential
future state of this situation (AR) and links between AR correspond to the actions
to be carried out and/or the conditions to be complied with to go from one AR to the
next one. Each AR is associated with two parameters indicating the potential risk
and the time saving to pass from the previous AR to this new one. In some cases,
these parameters come from default values (e.g. starting to cross the junction when
the lights are amber presents systematically more risk than stopping, but allow the
driver to save time). But generally, they depend on the current driving conditions
(e.g. according to visibility, the configuration of the crossroads and the traffic
conditions, the risk will be more or less important). Anticipation agent is in charge
to assess the value of these parameters for each new temporal derivation (i.e. AR
generation). This assessment is partly based on the relative zone conflicts between
the own driver's car, on the one hand, and other road users, on the other (see
Fig. 18.4). Furthermore, as the elaboration of the AR-tree progresses, anticipation
computes the overall value of these two parameters for the whole branch that
integrate the new derived AR (the whole sequence of AR starting from the RTC-
root). These 'local' (from one AR to another one) or 'global' values (for the whole
branch) will be then used by the decision agent to select the best action (or sequence
of elementary actions) in the current context.

18.2.2.4 The Knowledge Bases (KB) of the Tactical Module

Several research carried out at INRETS (e.g. Van Elslande, 1992; Dubois et aI.,
1993; Bellet, 1998) on drivers knowledge modelling have permitted to identify two
 18. Cognitive Modelling and Computational Simulation 327

major types of driving knowledge: declarative knowledge describing the road con-
text and operative schemas including procedural knowledge (i.e. practical know-
how) required to drive the car in the road environment. Regarding COSMODRIVE,
three specific KB are implemented at the tactical level: the road environment cat-
egories KB, the familiar place KB and the driving frame KB. These different
knowledge bases are not independent from each other. Strongly interconnected,
they constitute a global network handled by two cognitive agents specialised in
information retrieval: categorisation and place recognition. The road environment
categories KB is a typology of the different types of driving environment likely to
be encountered by the driver. This KB is hierarchically structured (Bellet, 1998).
At the top of the hierarchy can be found very generic categories corresponding to
classes of driving contexts (urban, country, highway). On descending the hierarchy,
the knowledge becomes more and more specialised (e.g. urban junction, access
lane), until reaching precise categorisations of the driving context (e.g. 'crossroads
with traffic lights in city centre'). The familiar places KB holds specific knowl-
edge of the road sites familiar to the driver. This KB is a direct extension of the
environment categories base: each familiar place constitutes a specific instance of
one of the classes represented by the road categories. Lastly, the driving frame KB
holds the driver's operative knowledge, as described earlier in this paper. Once
activated and instantiated with reality (matching with objective characteristics of
the infrastructure and integration of dynamic events), the driving frame becomes
the CTR, and this mental model will be used for driving the car.

18.2.2.5 The Cognitive Processes of the Tactical Module

The central aim of this module is to simulate mental model of a road scene (CTR)
elaboration in the human cognitive system. As mentioned above, this represen-
tation is built by matching a driving frame with perceptive information extracted
from the driving environment. Integration of the perceptive information can be
done according to two distinct modes. Either the information is actively sought
by the driver (e.g. expectations of the probable event occurrence in such area of
space), or the information is particularly salient and catches driver's attention (like
a red light at night, on a deserted road). The first type of information integration
is called top-down and is performed in COSMODRIVE by the perceptive explo-
ration process. Indeed, this agent receives all the perceptive exploration queries
from all the model's cognitive agents. According to their level of priority, per-
ceptive exploration defines the visual strategies permitting the best response to
these queries, then it positions the glance at the road scene. This leads to a change
in the buffers of the perception module, and as soon as the information sought
is found, it is sent to the requesting process(es). The second type of information
integration is bottom-up and performed by the cognitive integration process. This
can be likened to an array of filters that allows only the information of the percep-
tion module having greater physical magnitudes than certain thresholds (e.g. size,
speed of movement, colour, shape, etc.) to pass into the tactical representation. The
filters used as detection thresholds can vary according to the spatial position of the
328 Bellet et al.

object in relation to a fixed point (e.g. distance, eccentricity), the characteristics

of the environment (e.g. ambient luminosity) or those of the sensorial receivers
(e.g. visual pathologies). These mechanisms, and the whole Perception module,
are described in Mayenobe et al. (2004). Beyond perception module, five cognitive
agents co-operate at the tactical level: (1) categorisation, (2) place recognition, (3)
tactical representation generator, (4) anticipation and (5) decision. These agents
run in parallel and they are potential users of cognitive resources.

18.2.2.5.1 Categorisation
The main function of this process is to provide a more or less precise categorisation
of the road environment into which the driver is going. Furthermore, this agent
simulates the retrieval processing of a relevant driving frame for car driving. To do
this, categorisation attempts to match, on the one hand, the information contained
in the hierarchy of road environment categories KB and, on the other, the data
available in the fact list of the CSB. If the facts list does not contain any of the
discriminatory criteria permitting the selection of a category, categorisation sends a
queries to the perceptive exploration agent (perception module) in order to extract
new data from the road scene. As soon as a category of road environments is
identified, categorisation considers the list of goals to be attained, then it activates
one of the frames (permitting to reach this goal) linked with this category (driving
frame KB). The chosen frame is then transferred in the CSB, in place of the
previous CTR.

18.2.2.5.2 The Place Recognition Process

As categorisation, the place recognition (PR) agent intervenes in recovering a
driving frame from LTM. In this perspective, PR examines the CSB and attempts
to match the facts it contains with the characteristics of one of the many places
stored in the familiar places KB. As soon as a familiar place has been identified,
PR examines the goals list, selects one of the driving frames associated with this
familiar place, then it assures its transfer in the CSB. The tactical representation
generator (TRG) agent can then use this representation 'as is' with the aim of ori-
enting the driving activity in this particular place. Although the generic knowledge
recovered by categorisation requires adaptation in order for it to fit with reality,
the familiar place frames are very specific from the outset. Instantiation procedure
implemented by TRG is therefore reduced (limited to only dynamic events) and
its cognitive cost decreased.

18.2.2.5.3 The Tactical Representations Generator Process

This agent is the'orchestral conductor' of the tactical module. The main cognitive
tasks carried out by TRG are (1) the driving frame instantiation by integration of
the current situation characteristics, (2) integration and interpretation of new data
coming from the perception module, (3) to verify the CTR validity (e.g. has the
situation changed; is the goal pursued still valid?) and, if necessary, participate in
changing it, (4) activation the decision agent in order to determine what driving
 18. Cognitive Modelling and Computational Simulation 329

( Messages (e.g., ' factCW, X, Y,Z)' false =>W iswroog) )

I

, - - - - - -Goa ls L 1ST
Global Goal : saY. a_
Local Goal (current) : TurnLft (lIJ

FtIcls L 1ST
fact(obj eet. p osition . dynamic . chara cteristics)
foct(tnffic lights. tollC _ <>< 1. type_, tatic. color : 7)
foct(objeet: W. position: X. dynamic: Y. chlU1lCt : Z)

I
CURRENT T ACl7CAJ. R EPR~NTmON (CTR)
1) Frame : ~ban Cf'0661'O(Jds ""tir tniffi c lights: 1I."
2) Validity limit s : - facl( traflic li.ghts. static. exl . _)
- facl( W. X. Y. Z)
3) Infrastruc OJre :

KNOWLEDGE
BASCS

exl
traffic lights Current Position : Final Position :
color ? entry of Z 1. speed = 40 exit ofZ l .
kmlb. not any vehicle ahead speed = 25 kmIh
CurrentPosition :
lWHoo---EnJry into Zl,
~ei=40~. ::::/'::::::
,:! om
. : :::::::;:
:::::::
i ioNAL M ODULE,:"'::::'t,,,
:::: .:::;.;: .
: ::: : ::: :: : : :::: : :: :.: :: : : : : : :: : : :::: : :: ~ : :. : .' .'
CVIlRENT S TAn: BUCKBOARD

FIGURE 18.8. The TRG agent (tactical representations generator).

action should be taken, (5) sending this action to the operational module for its
effective implementation and (6) checking that the chosen action provides the
expected effects. Figure 18.8 illustrates the general activity of TRG in the tum
left (TL) situation at an urban crossroads as observed on a real site. At moment
to, the driver enters the junction (initial state of Z 1, speed 40 km/h). During the
seconds preceding to, the categorisation agent has selected a generic drivingframe
'crossroads with traffic lights: turn-on-the-left', then it transfers this frame to the
CS Blackboard. From this time, TRG task is to instantiate this frame with reality.
At this level, TRG must match the generic infrastructure contained in the frame
with the real road environment. Dynamic objects present in the road scene are then
integrated in the frame. At the end of this instantiation procedure, COSMODRIVE
has a CTR corresponding to its own mental model of the world at this time to. If
nothing contradicts it, this CTR will guide the driver 's actions until the frame is
totally performed (exit of crossroads at Z4), according to some updating in order
(a) to integrate new facts (TRG constantly consults the facts list for this purpose)
and (b) take into account the progression of the vehicle's current position as it
moves. Furthermore, a TRG verification procedure is in charge to assess the CRT
validity. To carry this out, TRG verify if the changes observed in the facts list do
not contradict the frame's validity limits (e.g. what has initially been identified as
a traffic light is in fact a shop sign) . It will also ensure that the goal pursued can
330 Bellet et al.

actually be reached (e.g. diversion, prohibiting a left tum). If this occurs, TRG
sends a message to categorisation in order to select a new frame, more adequate
with the new characteristics of the situation.
Now, let us return to moment to, and examine in more detail the TRG com-
putations in the seconds that follow (Fig. 18.8). At this time, TRG has a mental
model of the environment represented by the frame 'tum left at crossroads with
traffic lights' instantiated with the real situation. Taking into account knowledge
stored in this frame (perceptive exploration zones linked with the entry into Z 1),
the first COSMODRIVE reasoning will concern the traffic lights colour in exl.
As the information is not available in the facts list, TRG generates a query to per-
ceptive exploration (writing in the blackboard associated with this process). The
query taken into account, perceptive exploration positions the glance in ex1. Once
the road scene has been explored, the information gathered is transmitted to TRG
(message-response indicating that the light is green). TRG can then instantiate the
frame (light colour ex 1 == green) and choose one of two final positions associated
with Z 1: arrive at the exit of Z 1 with a speed of 25 km/h if the light is green and
at 0 kmIh if it is red (amber light although has also been taken into account). TRG
is thus able to transmit part of the frame to the operational module (characteris-
tics of Z 1, current position, expected final position and the conditions required
to attain this goal, e.g. the light remains green). Then, the operational module
determines the elementary actions to be taken (e.g. on the peddles and steering
wheel) to change from 40 to 25 km/h along the ZI straight length. This procedure
only partially describes COSMODRIVE reasoning for decision making. In fact it
is also equipped with anticipatory capacities: TRG usually does not need to take
decision itself, but only validates the decisions previously taken by anticipation
and decision agents.

18.2.2.5.4 The Anticipation Process

This agent's task is to predict changes in the current situation so as to orient driving
behaviour. Anticipation mechanism is essential in order to overpass the reactive
mode generally proposed by behavioural models. From a cognitive point of view,
anticipation permits the human drivers to project themselves into the future. It is
therefore possible to predict events and, more generally, to self prepare for any
situational change before it occurs. Mental simulation is also used to assess the
potential effects of an action and, on this basis, to compare this action with some
of its alternatives in view to determining the most relevant behaviour in the current
context. Anticipation is particularly essential in a dynamic situation. Without any
capacity for anticipation, the driver's chances of survival should be limited! The
greater the pressure of time becomes, the less time there is to act in situation, and
the more anticipation plays a decisive role. At the level of COSMODRIVE, the an-
ticipation process has the task of deriving, on the basis of the CTR, possible future
states in the driving environment. Each derived state constitutes an AR. All the
ARs generated are organised in the form of a tree whose nodes are constituted by
different ARs and whose inter-AR links correspond to actions to be taken and/or
 18. Cognitive Modelling and Computational Simulation 331

conditions required to go from one AR to the next. The anticipation procedure

starts by the transfer of the CTR in the AB. This will constitute the root of the ARs
tree. Once this transfer has been performed, anticipation successively examines
the different movement zones composing the driving path of the frame. It should
be recalled that a movement zone is characterised by an initial state (the vehicle's
position and speed when entering the zone), by a goal state (position and speed
when leaving the zone), and by an elementary action to be taken to reach this goal
state (on the tactical level, this action is limited to the command, 'attain goal state
in this zone' which will be intimated to the operational module). This action is
associated with conditions that cover events liable to occur in certain sections of
the road environment (movement zones and perceptive exploration zones). Dur-
ing this examination, anticipation 'virtually executes' the frame. This amounts to
'taking stock': each zone is considered, the goal states are defined and the nec-
essary conditions are explored (transmission of queries to perceptive exploration
in order to check the facts corresponding to these conditions). Nonetheless, this
'deployment' of the frame can be so costly in cognitive terms that, according to
the available time and resources, it will not be possible to derive all the ARs. In
this context, anticipation will give priority to the movement zone(s) immediately
consecutive to the current zone. Beyond these, the decision agent determines the
better derivation strategy according to its own requirements.

18.2.2.5.5 The Decision Process

This agent's task consists in examining the different ARs contained in the AB, in
making a selection from among these and in instantiating the CTR of the CSB
as a consequence. The TRG agent then transfers the actions associated with these
different ARs to the operational module, in order to implement them. More specifi-
cally, the decision-making process supposes (1) to consider the AB with the aim to
determine what branch of the AR tree is most adapted to the present context, (2) to
select the first AR of the branch (that immediately succeeding the current zone
already taken in hand by the operational module), (3) to instantiate the correspond-
ing movement zone of the CTR contained in the AB, (4) to wait for the message
from TRG indicating that this solution has been transmitted to the operational
module, (5) to restart the same procedure for the following zone and (6) so forth
until reaching the last zone of the frame. Furthermore, the decision agent does not
simply wait for solutions proposed by anticipation. It also plays an active role in
the AR derivation procedure. Decision permanently supervises the work done by
anticipation: Each new derived AR is examined with respect to decision-making
criteria (such as risk run and saving or loss of time) and, according to whether
these criteria are satisfied or not, decision determines the best derivation strategy
to be adopted. Insofar as the criteria of the moment are conformed to, an 'in-depth
first' research strategy will be given priority in order to result in one of the frame
goal states as soon as possible. Inversely, if the derived AR proves itself to be
incompatible with the decision-making criteria taken into account at this moment
332 Bellet et al.

(risk judged too high, for example), decision will choose a 'wide first' strategy in
order to examine other ARs liable to offer a more appropriate solution.

18.3 Methodology to Study Driver's Situation Awareness

As previously described, COSMODRIVE provides a theoretical framework to
analyse and simulate the cognitive processes involved in drivers' mental model
elaboration. This part of the chapter presents a specific methodology developed at
LESCOT (Bailly et aI., 2003; Bailly, 2004) in order to experimentally study the
'quality' of these internal models (adequacy with the reality), according to several
sources of variation like driver's experience (experienced versus novice drivers
comparison) and the quantity of cognitive resources available and/or the attention
sharing effect due to the impact of a secondary task of mental arithmetic. Beyond
these fundamental objectives, the aim was also to design a tool able to raise drivers'
awareness of a secondary task impact on driving performance.

18.3.1 Main Hypothesis

Two main hypotheses have been investigated in this part of the research program.
The first one suggests a better performance for experienced drivers (EDs) than for
novices due to the benefit given by driving knowledge. The second one postulates
a negative impact of a secondary task on the experiment performance. Concerning
the potential impact of driving experience, several studies have shown better per-
formances of EDs in terms of visual search, information processing, mental model
elaboration and driving situation understanding (Mourant and Rockwell, 1972;
Chapman and Underwood, 1998; Bellet, 1998; Crundall et aI., 1999). According
to the COSMODRIVE framework, it is expected that the driving frames used by
expert drivers allow them to have better perceptive explorations of the road scene
(i.e. focused on the most relevant information), increase information processing
performances and, consequently, provide a more adequate mental model of the
current driving situation (i.e. better understanding and awareness of what is hap-
pening on the road). Concerning the potential effect of a secondary task, several
studies have shown that parallel activities while driving (e.g. telephoning) have
a negative impact in terms of reaction time, visual strategies, situation awareness
and, more generally, risk of accident (Lamble et aI., 1999; Recarte and Nunes,
2000). From this point of view, it is expected that participants' mental model will
be potentially impaired by the mental arithmetic task. Beyond that this research
aims to determine more precisely the nature of this effect on the CTR (as described
in COSMODRIVE model). Two main questions will be more particularly consid-
ered: Which kind of information is more 'robust' in terms of lack of cognitive
resources impact (event occurrences versus infrastructure components like road
signs)? Are there any differences according to the distance of the element consid-
ered (e.g. is far versus nearby information more or less impaired by the secondary
task)? The last aspect of this experiment is focused on potential links between
 18. Cognitive Modelling and Computational Simulation 333

these two sources of variation (driving experience, on one side, and secondary
task impact, on the other side). This research aims also to compare the accuracy
of the CTR, according to a combined effect of these two dimensions.

18.3.2 Methodology
A specific methodology has been developed in order to study these questions
(Bailly et al., 2003; Bailly, 2004). The experimental plan comprises 40 video
sequences of driving situations. Participants are invited to look at these video films
on a TV screen. At the end of each sequence, the video suddenly stops. Then, the
last picture of the video film appears on the TV and subjects have (a) to determine
if something has changed or not and (b) to indicate the nature and the location
of the detected modification(s). Modifications can occur on events (e.g. add or
suppress car) or on road infrastructure components (e.g. add or suppress road
signs or change traffic lights colours). Forty participants (20 novices and 20 EDs)
took part in this study. During the first phase of the experience, the subjects' task is
limited to looking at 20 video sequences (simple task condition). During the second
phase, video observation is done with an additional task (mental arithmetic; dual
task condition). At this level, the aim is to analyse the impact of a secondary task
on the tactical representation content and the attention sharing strategies required
through lack of cognitive resources. Note that the 40 video sequences have been
divided into two series. According to a turning experimental plan, each of the two
series has been equally used in the simple task (ST) and in the dual task (DT)
conditions.

18.3.3 Main Results

Results obtained reveal several things. Firstly, the modification detection is strongly
dependent on the driving experience. Indeed, if we consider the global performance
(including both experimental conditions, ST and DT), EDs obtained better results
than novices (68.2% versus 50.6% of right detection) (Fig. 18.9). If we consider
driving experience effect for, respectively, each of the two experimental conditions,
this difference is significant for the ST (74.9% for ED versus 58.5% for novices)
as well as for the DT condition (61.5% versus 42.7%). From another point of view,
the lack of cognitive resources has a negative effect on the detection performance.
For EDs, the detection performance decreases from 74.9% (ST) to 61.5% (DT)
if cognitive resources are allocated to an additional task. The same effect is ob-
served for novice drivers (58.5% versus 42.7%). These results clearly demonstrate
that cognitive resources are required to build a reliable mental model of the driv-
ing situation: The DT negative impact on mental model quality is systematically
significant and concerns all drivers.
Concerning driving experience and DT effects, according to the nature of modi-
fications (events versus infrastructure), Fig. 18.10 shows that EDs are always better
than novices of detecting all types of modification (for infrastructure as well as
event).
334 Bellet et al.

100 , -- - - - - - - - - - - - - - - - - ---,
90 + - - - - - - - - - - - - - - - - - - - - - - l
~ 80 -t-- - - - - - - - - - - - - - - - - - - ---\
.2 70 -t--== - - - - - - --j
ti
~ 60
c 50
1:
40
a:'0
Cl

30
<ft 20
10
o
global simpl e task dual task

FIGURE 18.9. Percentage of right detection according to the experimental conditions (ST
versus DT).

For the ST condition,EDs detected 79.1% of the road infrastructurecomponent

modifications, whereas novices detected only 60.9%. Concerning event modi-
fications, EDs obtained 70% of right detection, whereas novices obtained only
55.8%. The same tendency is observed for the DT condition. EDs detected 57.3%
of road infrastructure modifications, whereas novices detected only 36.4%. For
events modifications, EDs' performance is 66.3% and novices' performance is
50%. If we consider, for each population of drivers (ED versus novices), the sec-
ondary task impact according to the nature of modifications, the results reveals
a difference between event versus infrastructure. On the one hand, the secondary
task has a significant effect on the infrastructure modification detection for all

100
90

til
80
c:
.2 70
ti
Ql 60
Qi
c 50
1:
a:
Cl
40
'0 30
-:!!.
0
20
10
0
Infrastructure Events

o Experienced ST Experienc ed DT 0 Novices ST Novices DT

FIGURE 18.10. Percentage of right detection according to thenature of themodification.

 18. CognitiveModelling and Computational Simulation 335

[15-25m] [25-50m] [+50m]

_______ Exp. DT -tr- Nov. ST -O- Nov. DT

FIG URE 18.11. Percentage of right detection accordingto the distance of modifications.

the drivers (respectively, 79.1 % versus 57.3 % for ED and 60.9 % versus 36.4 for
novice drivers ). On the other hand , no significant effect was found concerning
event modifications according to the ST versus DT condition (even if the perfor-
mance is globally better for ST condition). The result is the same for EDs (70%
in ST versus 66.3% in DT, but khi , = 0.594, p = 0.441 ) and for novice drivers
(55.8% in ST versus 50 % in DT, but khi- = 1.278, p = 0.258). In other words ,
event modification detection is clearl y more ' robust' (i.e. lesser effect of the DT
condition) than infrastructure components modification detection.
The last comparison presented herein concerns the distance of the modified
component. Four distances have been defined: 0 to 15 m (nearby zone), 15 to
25, 25 to 50, and beyond 50 m (jar zone). Performances for each distance are
compared according to the driver s' experience and to the secondary task effect (Fig.
18.11). In ST conditions, EDs obtain better results than novice s for modifications
concerning the nearby zone (81.3 % versus 60%) and also for modifications in the
15 to 25 m zone (87 .1% versu s 67.1%). Beyond 25 m, differences between ED
and novices are not significant. For the DT conditions, differences between experts
and novices are significant for three of the four distances con sidered. Performances
are, respectively, 67 .5% (expert) versus 51 .3% (novices) for the nearby zone (0
to 15 m), 67.1 % versus 42 .9% for the 15 to 25 m zone , and 70% versus 41.4%
for the 25 to 50 m zone. No significant effect is found for the far zone. If we now
con sider, for each population of drivers, the secondary task impa ct according to the
distance of modifications, the result s reveal some relevant differen ces. For EDs, a
DT significant effect occurs in the nearby zone (8 1.3% versus 67.5 %) and for the
15 to 25 m zone (87.1% versus 67. 1%). However, for novices drivers, DT impact
336 Bellet et al.

concerns the 15 to 25 m zone (67.1 % versus 42.9%) and the 25 to 50 m zone (64.3%
versus 41.4%). For all the drivers, no significant difference is found beyond 50 m,
probably because the performances were already weak in ST conditions. The other
main relevant 'non-significant' effects of the DT conditions concern specifically
(a) only novice drivers for the 0 to 15 m zone and (b) only EDs for the 25 to 50 m
zone.

18.3.4 Discussion and Conclusion Concerning Experimental

Study ofDrivers Situation Awareness
At the global level, results obtained with this methodology confirm initial hypothe-
ses. Firstly, modification detection is better for EDs than for novices, whatever the
experimental conditions (ST and DT). Insofar as modifications concern only rele-
vant information for safe driving, it means that driving experience is a key element
for elaborating an accurate mental representation of the current driving situation
(i.e. better understanding and awareness of the driving context). Secondly, the in-
troduction of a secondary task (mental arithmetic) has a negative impact on the
mental model elaboration. This effect is significant for all drivers (novices as well
as EDs). Fundamentally, to carry out a mental arithmetic task, cognitive resources
are required. Consequently, other mental activities that require cognitive resources
(i.e. involving controlled processes) will be affected (i.e. cognitive resources and
attention sharing between the primary and the secondary task) (Schneider and
Schiffrin, 1977; Schneider and Chein, 2003). On the contrary, the DT does not
have impact on automatic processing. From this theoretical point of view, results
obtained in this experiment reveal several things. Firstly, mental model tactical
representation elaboration partly requires cognitive resources (i.e. mental model
building is not only based on automatic processing). Moreover, the DT effect is
proportionately more important for novices than for EDs. It means that driving
experience also has a positive effect in managing a secondary task and in sharing
attention during driving. Lastly, the DT impact allows us to reveal the level of
priority of information collected in the environment and integrated in the drivers'
mental model. Indeed, the DT condition considerably reduces the quantity of cog-
nitive resources available. Consequently, only the most relevant and salient pieces
of information will be taken into account. In terms of experimental test perfor-
mances, it means that the secondary task effects are inversely proportional to the
level of relevance and/or salience of information. In other words, 'non-significant
effects' of the DT condition (compared to the ST condition) will indicate a high
level of relevance and/or salience of this specific information. Concerning this
aspect, an interesting result has been obtain according to the distance, via the spe-
cific 'no significant effects' found, respectively, for each population of drivers. For
EDs, DT does not impact detection performance for the 25 to 50 m zone, whereas
for novices, DT have no effect for the nearby zone (0 to 15 m). This result reveals
a difference between novices and EDs concerning the main 'focal point' used for
visual scanning of the driving scene. For novices, perceptive explorations are first
 18. Cognitive Modelling and Computational Simulation 337

and foremo st focused on the nearby environment of the vehicle. For EDs, the focal
point is between 25 and 50 m ahead the car (between 2 and 3.5 s).

18.4 Some Experimental Results Simulation with

Cosmodrive
By considering this type of experimental results, it is possible to use COSMOD-
RIVE simulation to visualise drivers' mental model s of the road scene at different
times. Figure s 18.12 to 18.14 give some exa mples of mental models content, ac-
cording to driving experi ence and attention sharing (i.e. lack ofcog nitive resources)
effect s.
Figure 18.12 typically illustrates mental model differenc es between experienced
versus novice driver s, in ST conditions (i.e. all driver's attenti on is focu sed on the
road ). At this time of the drivin g situation, main dynamic significant event s (salient
and close) have been perceived and integrated by all the driver s in their mental rep-
resent ation . Nevertheless, only EDs have generally take into acco unt the car coming
in front of us, with the future aim to tum on the left. Accordin g to COSMODRIVE
model, it is possible to explain this result by anticipation age nt reasonin g throu gh
the turnin g-left frame deployment (i.e. ment al simulation of the driving situation
evolution). An event is more particularly significant if its trajectory interferes with
our own driver's trajectory. From this point of view, the opposite car will be par-
ticul arly important in the future. Nevertheless, novices' attention is essentially
focused on interactions with nearest events. Consequently, they generally neglect
this long-term event at this time. On the contrary, anticipation abilities of EDs
allow them to anticip ate the future ' turning-left' phase (i.e. mental examination of
the driving frame 'Z3' zone) for adequately plannin g their behaviour. As a result,
the majorit y ofEDs are generally aware, since the approaching phase (Z l) of this
incom ing car occ urrence and of the potenti al conflicts with it in the future .
Figure 18.13 illustrates the potent ial impact of attenti on sharing on mental mod-
els. For novice drivers, the lack of cog nitive resource s involves an attention focal-
isation on the nearby zone (i.e. vehicles beha viours on the left side), but nothing

FIGURE 18.12. Novice (view) versus experienced drivers (right) mental model in a ST
condition.
338 Bellet et al.

FIGURE 18.13. Novice (left) versus experienced (right) drivers' mental model in dual-task
condition .

is observe and known concerning the traffic flow in the intersection itself. On the
contrary, EDs focused as a priority the residual attention on the 25 to 50 m zone. If
some secondary events are potentially missing, experienced drivers have generally
integrated at this time the most significant events (i.e. car coming from the opposite
direction) in relation with their own "left-turn" goal.
Lastly, Fig. 18.14 illustrates the COSMODRIVE simulation interest in order
to dynamically visualise mental models evolution, during the whole approaching
phase of the crossroad. If we consider, for example, the previous novices' mental
model in DT condition, drivers' attention will gradually be focused on the area
ahead as they progress on the road and the traffic flow will be finally considered
(and partly integrated) a few seconds latter.

FIGURE 18.14. Example of simulation of a novices' mental model evolution during the
whole crossroad approaching phase (zone Z I of the driving frame) .
 18. Cognitive Modelling and Computational Simulation 339

18.5 Conclusion and Perspectives: From Behaviours

to Mental Model

Several recent researches, carried out at INRETS-LESCOT, have been presented in

this paper. The common point of these works is the drivers' mental model study (i.e.
internal representation of the current driving situation), with the aim to simulate
this cognitive component on computer. The central part of this human modelling
program is the COSMODRIVE described in the third part, by considering (a) its
cognitive architecture, (b) the formalism used to represent driving knowledge and
simulate mental models and (c) the cognitive processes involved at the tactical
level of the driving activity. The fourth part of the paper focused on a specific
methodology developed at LESCOT in order to experimentally study the 'quality'
of these mental models (i.e. its adequacy with the information available in the
road scene), according to several sources of variation (driver's experience and
cognitive resources available). We would like to conclude this paper by a short
presentation of a starting research in the field of Humanist Network of Excellence,
with the aim to define a theoretical and methodological framework to carry out a
cognitive analysis of the car driving activity. From a practical point of view, the
goal is to develop models, methods and cognitive engineering tools for' ecological'
data (collected in real driving conditions) processing and analysis in order to infer
driver's situation awareness (i.e. mental model) from driving activity 'traces'.
Concerning drivers behaviour modelling, the research will use the 'driving frame'
approach defined at LESCOT. Indeed, driving frames formalism can be used for
modelling the driving activity as well as the driver's mental models. This new
research will aim to define methods and develop computational tools in order to
'abstract' the driving frames from objective data including (a) driver's behaviours,
(b) vehicle state and (c) environmental conditions.
Concerning artificial intelligence and cognitive engineering tools, the research
will use the MUSETTE (modelling uses and tasksfor tracing experience) methods
(Fig. 18.15) developed at LIRIS (Champin et aI., 2004). Figure 18.16 illustrates

FIGURE 18.15. 'Primitive trace' encoding:

1 The MUSETTE theory provides a method for encoding traces of human behaviours.
The first level of trace, called rough trace, is made up of a succession of observables,
called objects of interest (01). The second level of trace, called primitive trace, is a suc-
cession of states and transitions embedding the Ols. Ols are of three kinds: 'entities'
(static facts) allowing description of states, 'events' allowing description of transitions
and 'relations' from entities or events to each other. In Fig. 18.15, states are represented
by circles, transitions by rectangles, entities are labelled En, events are labelled Ev and
340 Bellet et al.

(a) Normal

Zl Z2 Z3

rnI Infrastluctun! zones:

(b) Surprised ZI :Approach
Z2:ClOssing
Z3:Clearing
Zl Z2 Z3

FIGURE 18.16. Normal versus surprised operational sub-schema.'

how it is possible to connect a low-level trace of the driving activity (i.e. opera-
tional sub-schemas) to the upper descriptive level embodied by the tactical driving
frames. This work falls under the LESCOT's effort to better understand the car
driving activity. Within this new framework, the possibility of automatically as-
sess the driver's situation awareness from driving behaviours could have multiple
uses such as to get a better understanding of the accidents or to identify needs or
possibilities of assistance. In a long-term perspective, it could also be a first step
towards adaptive technologies based on a mental model sharing between, on one
side, the human driver and, on the other, driving aids (Bellet et aI., 2003).
Providing assistance that is adapted to the current driver's needs and to the spe-
cific characteristics of the driving situation of the moment is the aim of adaptive
technologies. The objective is to head towards the contextualisation of driving aids.
It is easy to imagine that driver's needs may be different depending on the type

relations are labelled R. The cutting into states and transitions is useful to identify episodes,
which consist of passing from one state to another through a succession of intermediate
states.
2 The creation of the primitive traces led us to widen a tactical driving frame to a finer
granularity level. This level is called operational sub-schema. This representation allows us
to describe tiny differences between situations. Figure 18.16shows how we can represent a
'surprised driver' sub-schema that is characterised by an initial 'Cruise' state that encroaches
on the approach zone (Zl ) of the infrastructure, followed by a 'strongly decelerate' state
characterised by a value of deceleration higher than in the normal sub-schema. From future
developments we want to implement search functionalities, which enable us to find patterns
in the primitive trace likely to match with a tactical driving frame, according to specified
criteria (Georgeon et aI., 2005) .
 18. Cognitive Modelling and Computational Simulation 341

of the driving task (e.g. monotonous driving on motorways, going through roads
junctions in urban areas, overtaking), depending on the driving conditions (e.g.
traffic density, weather conditions), or depending on the driver's permanent (e.g.
age) or momentary capacities (e.g. cognitive resources available, state of stress,
fatigue). Contextualising aid involves developing technologies that are 'open to
the world', that is to say capable to assess - in real time - the driver's activities and
all the requirements dictated by driving conditions. This must all be done to (a)
determine if it is necessary or not to assist the driver in the present situation and (b)
identify the kind of aid required in this specific context and the form it should take
(giving information to the driver, warning him of a danger, or taking control of the
vehicle). We can also imagine using this kind of analysis capacity for diagnosing
human errors: Is the driver's behaviour appropriate according to the characteristics
of the driving situation? Alternatively, does the driver's behaviour present certain
risks not taken into account by human awareness? Several researches have been
done at LESCOT concerning adaptive technologies (see Bellet et al., 2003), with
the aim to manage on-board information or to design cooperative collision avoid-
ance system. Nevertheless the previous works were based on behavioural analysis
methods, but not on a cognitive simulation model. This new research will explore
this possibility by using some part of COSMODRIVE model. It could be a first step
to bring man-machine cooperation closer to man-man cooperation situations in
which two human agents (the driver and the co-driver) co-ordinate their efforts to
carry out the driving task together. In this context, each partner mentally imagines
the activities of the other partner (i.e. activities he does and those of which he is in
charge) and uses this internal model for interacting with him in an appropriate way.
In terms of man-machine cooperation, this means 'symetrising' the man-machine
relationship (Amalberti and Deblon, 1992) and developing technology capable to
represent itself the human partner activity (by using a more or less complex internal
model) like the human partner imagines (through a more or less true mental model)
the technological system in charge to assist him or her (i.e. what it is supposed to
do, what it really does and, if needs be, the way it goes about doing it).

References
Allen, T.M., Lunenfield, H. and Alexander, G.J. (1971). Driver information needs. Highway
Research Record, 366, 102-115.
Amalberti, R. and Deblon, F. (1992). Cognitive modelling of fighter aircraft's control pro-
cess: A step towards intelligent onboard assistance system. International Journal of
Man-Machine Studies, 36, 639-671.
Amidi, O. (1990). Integrated mobile robot control. Technical Report CMU-RI-TR-90-17.
Carnegie Mellon University Robotics Institute.
Bailly, B. (2004). Conscience de la situation des conducteurs: Aspects fondamentaux,
methodes, et applications pour la formation des conducteurs. PhD Thesis, Universite
Lyon 2.
Bailly, B., Bellet, T. and Goupil, C. (2003). Driver's mental representations: Experimental
study and training perspectives. In Proceedings of the 1st International Conference on
Driver Behaviour and Training, Stratford-upon-Avon, England. CD-ROM, 8 pp.
342 Bellet et al.

Bellet, T. (1998). Modelisation et simulation cognitive de l'operateur humain: une appli-

a
cation La conduite automobile. PhD Thesis, Universite Paris V.
Bellet, T. and Tattegrain-Veste, H. (1999). A framework for representing driving knowledge.
Intenational Journal of Cognitive Ergonomics, 3(1), 37-49.
Bellet, T. and Tattegrain-Veste, H. (2003). COSMODRIVE: un modele de simulation cog-
nitive du conducteur automobile. In lC. Sperandio and M. Wolf (Eds.). Formalismes de
modelisation pour l'analyse du travail et l'ergonomie (pp. 77-110). Presses Universi-
taires de France, Paris.
Bellet, T., Tattegrain-Veste, H., Chapon, A., Bruyas, M.P., Pachiaudi, G., Deleurence, P. and
Guilhon, V. (2003). Ingenierie cognitive dans le contexte de l'assistance a la conduite
automobile. In G. Boy (Ed.). L'Ingenierie Cognitive: IBM et Cognition. Hermes, Paris.
Cacciabue, P.C., Decortis, F., Drozdowicz, B., Masson, M. and Nordvik, lP. (1992).
COSIMO: A cognitive simulation model of human decision making and behaviour in
accident management of complex plants. IEEE Transaction on Systems, Man and Cyber-
netics, IEEE-SMC, 22(5),1058-1074.
Champin, P.-A., Prie, Y. and Mille, A. (2004). Musette: A framework for knowledge capture
from experience. In Proceedings of EGC04, Clermont Ferrand, France.
Chapman, P.R. and Underwood, G. (1998). Visual search of dynamic scenes: Event types
and the role of experience in viewing driving simulation. In G. Underwood (Ed.). Eye
Guidance in Reading and Scene Perception. Elsevier, North Holland.
Crundall, D., Underwood, G. and Chapman, P.(1999). Driving experience and the functional
field of view. Perception, 28, 1075-1087.
Dubois, D., Fleury, D. and Mazet, C. (1993). Representations categorielles: perception et/ou
action? Contribution apartir d'une analyse des situations routieres, In A. Weill-Fassina,
P. Rabardel and D. Dubois (Eds.). Representations pour l'action (pp. 79-93). Octares
Editions, Toulouse.
Endsley, M.R. (1995). Toward a theory of situation awareness in dynamic systems. Human
Factors, 37(1), 32-64.
Georgeon, 0., Bellet, T., Mille, A., Lettisserand, D. and Martin, R. (2005). Driver behaviour
modelling and cognitive tools development in order to assess driver situation awareness.
In Proceedings of HUMANIST Workshop on Driver Modelling, Ispra, Italy, 25-27 May
2005.
Gibson, J.1. and Crooks, L.E. (1938). A theoretical field-analysis of automobile driving.
American Journal of Psychology, 51,453-471.
Hall, E.T. (1966). The Hidden Dimension. Doubleday, Garden City.
Fuller, R.A. (1984). Conceptualization of driving behaviour as threat avoidance. Er-
gonomics, 27(11), 1139-1155.
Kontaratos, N.A. (1974). A system analysis of the problem of road causalities in the United
State. Accident Analysis and Prevention, 6, 223-241.
Krajzewicz, D., Hiihne, R.D. and Wagner, P. (2004). A car driver cognition model. In
Proceedings of ITS Safety and Security 2004, Miami Beach, FL, 24-25 March 2004.
Larnble, D., Kauranen, T., Laakso, M. and Summala, H. (1999). Cognitive load and detection
thresolds in car following situations: Safety implications for using mobile (cellular)
telephones while driving. Accident Analysis and Prevention, 31, 617-623.
Levison, W.R. (1993). A simulation model for driver's use of in-vehicle information sys-
tems. Transportation Research Record, 1403,7-13.
Mayenobe, P. (2004). Perception de l' environnement pour une gestion contextualisee de
La cooperation Homme-Machine. PhD Thesis, University Blaise Pascal de Clermont-
Ferrand.
 18. Cognitive Modelling and Computational Simulation 343

Mayenobe, P., Blanc, C., Trassoudaine, L., Bellet, T. and TattegrainVeste, H. (2003). Active
perception tasks driven by a cognitive simulation model. In Proceedings of the IEEE
Intelligent Vehicles Symposium: IV-2003, Columbus, OH (pp. 378-382).
Mayenobe, P., Trassoudaine, L., Bellet, T. and TattegrainVeste, H. (2002). Cognitive sim-
ulation of the driver and cooperative driving assistances. In Proceedings of the IEEE
Intelligent Vehicles Symposium: IV-2002, Versailles, June 17-21,2002 (pp. 265-271).
Mc Knight, A.J. and Adams, B.B. (1970). Driver Education Task Analysis. Vol. I: Task
Descriptions. Human Resources Research Organisation HumRRO, Alexandria.
Mourant, R.R. and Rockwell, T.H. (1972). Strategies of visual search by novices and expe-
riences drivers. Human Factors, 14(4),325-335.
Michon, 1.A. (1985). A critical view of driver behavior model: What do we know, what
should we do? In Evans and R.C. Schwing (Eds.). Human Behavior and Traffic Safety
(pp. 485-520). Plenum Press, New York.
Minsky, M. (1975). A framework for representing knowledge. In P. Winston (Ed.). The
Psychology of computer vision (pp. 211-281). McGraw-Hill, New York.
Naatanen, R. and Summala, H. (1976). A model for the role of motivational factors in
driver's decision-making. Accident Analysis and Prevention, 6, 243-261.
Ochanine, V.A. (1977). Concept of operative image in engineering and general psychology.
In B.F. Lomov, V.F. Rubakhin and V.F. Venda (Eds.). Engineering Psychology. Science
Publisher, Moscow.
Otha, H. (1993). Individual differences in driving distance headway. In A.G. Gale (Ed.).
Vision in Vehicles IV. Elsevier, Netherlands.
Rasmussen, 1. (1983). Skills, rules and knowledge; Signals, signs and symbols and other
distinctions in human performance models. IEEE Transaction on Systems, Man and
Cybernetics, 13(3), 257-266.
Recarte, M.A. and Nunes, L.M. (2000). Effect of verbal and spatial-imagery task on eye
fixations while driving. Journal of Experimental Psychology: Applied, 6, 31-43.
Rumbaugh, 1., Blaha, 1., Eddy, F., Lorensen, W. and Premerlani, W. (1991). Object Oriented
Modeling and Design. Prentice Hall, Londres.
Salvucci, D.D., Boer, E.R. and Liu, A. (2001). Toward an integrated model of driver behavior
in a cognitive architecture. Transportation Research Record, 1779, 9-16.
Schneider, W. and Schiffrin, R.M. (1977). Controlled and automatic human information
processing I: Detection, search, and attention. Psychological Review, 84(1), 1-66.
Schneider, W. and Chein 1.M. (2003). Controlled and automatic processing: Behavior,
theory, and biological mechanisms. Cognitive Science, 27, 525-559.
Sukthankar, R. (1997). Situation Awareness for Tactical Driving. PhD Thesis, Carnegie
Mellon University.
Thorn, R. (1991). Predire n'est pas expliquer. Eshel, Paris.
Van der Molen, H.H. and Botticher, M.T. (1988). A hierarchical risk model for traffic
participants. Ergonomics, 31(4), 537-555.
Van Elslande, P. (1992). Les erreurs d'interpretation en conduite automobile: mauvaise
categorisation ou activation erronee de schernas? Intellectica, 15(3), 125-149.
Wierda, M. and Aasman, J. (1992). Seeing and driving: Computation, algorithms and im-
plementation. Traffic Research Centre, University of Groningen, Netherlands, Haren:
Verkeerskundig Studiecentrum, VK91-06.
Wilde, G.J.S. (1982). Critical issues in risk homeostasis theory. Risk Analysis, 2(4), 249-
258.
19
Simple Simulation of Driver
Performance for Prediction and Design
Analysis
P. C. CACCIABUE, C. RE, AND L. MACCHI

19.1 Introduction
19.1.1 Modelling Human Behaviour in Modern Technology
The ability to model and predict Human-Machine Interaction (HMI) with consis-
tent approaches is a crucial aspect of modern technological systems, where the con-
sideration for the presence of humans in control of systems is necessary at design
level as well as during production and implementation processes. A reliable and
realistic approach that enables to anticipate what will be "done" at different levels
of cognition and behaviour by a human being, enables the prevention of erroneous
or risky behaviour, as well the implementation of means of intervention exploiting
the power of modem technologies and the decision making skill of humans.
The literature of the last decades is rich with methods that focus on this subject.
A comprehensive and short review of methodological approaches and theoretical
constructs has been given by Moray (1997), covering the last 30 years of research
and development in the field of human factors applied to different domains and
industrial settings.
In modem systems, modelling human behaviour entails considering primarily
cognitive processes and system dynamic evolution, i.e., mental activities, resulting
from the time-dependent interaction of humans and machines. In this sense, a model
of human behaviour cannot be developed in total isolation and abstraction, but it
must be coupled and linked to a model of plant performance in order to develop
the simulation of a 'joint' human-machine system, also called 'Joint Cognitive
System' (Hollnagel and Woods, 1983; Hollnagel, 1993). Therefore any model of
human behaviour comes necessarily framed in a clear interaction context with the
associated technological system.
The formulation of models ofHMI is an evolutionary process whose origin lies in
the cybernetic paradigm of Wiener (1948), who developed the analogy between the
human operator and the servomechanism, and in the similarities between comput-
ers and brains pointed out by von Neumann (1958). These analogies have increased
the awareness of the closed-loop nature ofHMI, with the specific goal-oriented be-
haviour of the human controller. However, the most substantial theories on human

344
 19. Simulation of Driver Performance 345

behaviour derived from these first paradigms did not consider the predictive nature
of human behaviour, as all relevant human activities were practically associated
with manual control and direct interaction with physical phenomena. The inclusion
of the human element in the control loop of a process was thus a simple exercise
of mathematical consideration of action delays and parameters estimation.
The progress of technology towards supervisory control and automation re-
quired the formulation of much more complex models of human reasoning and
decision making processes, able to account primarily for cognitive rather than
manual activities. The need of simulating the man-plus-the-controlled-element was
then enunciated in the 1960s by McRuer and colleagues (1965). Other pioneering
research in this area, combined with the development of computer technologies,
inspired the first formulations of theoretical models of cognition (Neisser, 1967)
and, in the early 1970s, the metaphor of the operator as Information Processing
System (Newell and Simon, 1972).
Since then, a variety of paradigms of human behaviour have been developed
(Rouse, 1980; Rasmussen, 1983; Stassen et aI., 1990; Sheridan, 1992; Wickens,
1984, 2002), which aim to cover different levels of complexity and depth in repre-
senting mental processes, cognitive functions as well as behavioural performances.
These models vary quite substantially for many reasons. Certain models focus on
a specific domain of application. Other models pay attention primarily to cognitive
functions, such as decision making or diagnostic processes. Certain models con-
centrate on the role of operators and their interactions, such as supervisory control
process and teamwork.
This variety of models offers the analysts and designers of joint cognitive sys-
tems the possibility to choose the most suitable approach and paradigm to apply
for specific applications. However, in order to make the best selection in relation
to the problem at hand, the scope of application of the available methods and
models must be recognised. The existence of conceptual frameworks that enable
to identify boundaries and areas of consideration for modelling human behaviour
is discussed in detail in Chapter 2 of this book.

19.1.2 Modelling Drivers in the Automotive Context

Focusing on the problems associated with modelling driver behaviour, it is impor-
tant to realise that what may seem a simple domain to account for is instead much
more complex and varied than other similar environments with high levels of au-
tomation and complexity. The major peculiarities of the automotive environment
are associated with five specific aspects, namely: the Users of vehicles; the Tem-
poral Aspects of the HMI; the driving Working Context; the Social Environment
in which driving occurs; and finally, the type and mode of System Integration that
occurs with respect to the actual technical systems made available by different
manufacturers.

Users
A large variety of people can obtain a driver license. These include teenagers,
elderly and disabled persons. In other domains, the types of users are much less
346 Cacciabue, Re, and Macchi

variable. Moreover, drivers receive minimal initial training, often little or no

refresher training at all, and no training with specific new vehicles or devices.
In contrast, the users of plants and systems in other domains (e.g. pilots, air
traffic controllers, control room operators, medical staff, technicians and so forth)
may be highly selected, relatively homogenous, well trained and supervised or
monitored.
Temporal Demands
The temporal demands for in-vehicle warning practises are typically much more
time-critical than for other applications. For a potential crash situation, there may
be only a second or two between the recognition of conflict and the moment of
the impact. In contrast, emergency situations in other contexts may be measured
in tens of seconds or even minutes.
Working Context
Most technological domains are based on specific and dedicated workplace en-
vironments. This is not the case for automotive domain. This difference has
many implications in terms of acceptability of constraints and practices and as-
sumptions about the user. As an example, the acceptable level of intrusiveness
or annoyance and tolerance for false alarms in a car may be quite different than
in an aeroplane cockpit. In addition to this context 'internal' to the vehicle, there
is also a context that is 'external' to it, namely the roads and traffic. This is also a
very specific peculiarity of road transportation, as it represents a domain where
there is an extremely high density of traffic and the rules for traffic management
are almost completely left to the decision of drivers rather than a specific traffic
management authority.
Social Context
There is a social context in driving that is substantially different than other
environments. The presence of other passengers in the vehicle means that various
driver warnings may be public. Consumer acceptance of an in-vehicle-warning
device, or behavioural reactions to the warning, can be influenced in various
ways by the social context (e.g. embarrassments, feelings of competence, need
to appear daring or competent to peers). Social context is also present in other
domains, such as aviation, air traffic control, train driving, etc. However, in these
other environments, human behaviour is more affected by organisational factors
and national cultures than the local social context.
System Integration
In many applications and domains, the workplace is designed as an integrated
system so that the system requirements can be well defined from the beginning.
For vehicle applications many devices may be aftermarket add-ons, not inte-
grated into the original design of the vehicle. Then the environment is evolving
and it is not fully predictable, neither consistent from vehicle to vehicle.

This overall picture is further complicated by the massive presence, in modern

vehicles, of Advanced Driver Support Systems (ADAS) and In-Vehicle Informa-
tion Systems (IVIS), which add another level of complexity. For all these reasons,
the development of a model of driver behaviour is a very complex and difficult
 19.Simulation of Driver Performance 347

endeavour. However, the implementation of a model and associated simulation is a

necessary step forward for the improvement of the effectiveness of all new means
and systems that are designed and implemented in modern vehicles.

19.1.3 Use and Applications ofDriver Models

The models of drivers and operators in general can be utilised for many different
purposes. The goals of the users of models are the means for selecting amongst
the variety of proposed paradigms that can be found in the literature. The key for
choosing is therefore a set of criteria on which to compare the different paradigms
and approaches. One way to compare is to consider models ability to describe
performance or to evaluate motivational aspects. Another assessment perspective is
the capability to account for actual performances versus risk taking and perception.
A very important way to compare models is to consider their ability to describe
the dynamic interaction and predictive power versus their capability to describe
real-time behaviour in a realistic and coherent way.
In general, it is not possible to identify an architecture or a structure that is the best
in absolute, unless the generality of the approach is so wide that it looses site of po-
tential applications. What is instead very important is that the users of models have
a clear picture of the goals and objectives of the applications so as to select in the
range of paradigms the one that best suits their aims (Carsten, 2007, this volume).
In this paper, we focus on the development of a model and associated simulation
with the primary objective to be a predictive tool for studying possible Driver-
Vehicle-Environment (DVE) interactions dedicated to design and safety analyses.
Such type of tool is normally associated and compared with real-time applications
that can be utilised on board of vehicles, as they are usually integrated and stream-
lined version of the former model. This is considered an evolutionary process that
firstly realises a tool that is able to reproduce reasonable and verified behaviours
for design studies. These simulations are very useful for predicting behaviour and
DVE interactions that enable the designer and safety analysts to evaluate poten-
tial applications, drawbacks and advantages of new systems and design solutions,
without spending too much time and effort in performing field tests on expensive
prototypes and lengthy experiments. With a valuable model and simulation, it is
possible to assess potential errors and mishaps derived from various aspects of a
new tool, such as type of interface, location on board, effects on driver attitudes and
behaviour with respect to other systems, etc. Once such a simulation is tested and
validated, it is then possible that a simplified or streamlined version is developed
that can be utilised on board of vehicles for anticipating behaviour and avoid-
ing errors or improving information management by on-board communication
tools.
The simulation described in this paper concentrates on the first part of such
a development. It aims therefore at predicting DVE interactions in a sufficiently
accurate fashion to be utilised by designers and safety analysts for studies and
preliminary evaluations of HMls. The evolution of this approach into a real-time
tool is a natural follow up of the development, but is not discussed here.
348 Cacciabue, Re, and Macchi

19.1.4 Content ofthe Paper

In this paper, we will primarily describe the modelling architecture from the sim-
ulation perspective. We will concentrate on the algorithms that have been selected
to represent the driver behaviour in a form that can be implemented in a comput-
erised expression. Secondly, we will describe the different numerical formalisms
and the mathematical expressions utilised for simulating the dynamic interactions
with the vehicle and environment. We will then present a series of sample cases
of predictive interactions between drivers, vehicles and environments, which can
be used for design and safety assessment purposes. Finally, we will conclude with
some speculative consideration of the possible extension of this simulation in a
streamlined version that may be considered for real-time representation of driver
behaviour for in-vehicle systems implementation.

19.2 Simple Simulation of Driver Behaviour

19.2.1 Paradigm ofReference

The criteria for selection of the paradigm of reference for the development of this
simulation of driver behaviour have been the following:

1. The model must enable rapid assessments of DVE interactions with different
configurations so as to evaluate prototypes of different conceptual nature.
2. The dynamic interactions between driver, vehicle and environment need to be
included in the simulation. The vehicle and environment should offer dynamic
changing situations and interactivity with driver actions.
3. Cognitive aspects of behaviour and 'joint modelling' between humans and
vehicle-environment must be considered.
4. The model needs to account for behavioural adaptation to different types of
ADASs and possible emotional/attitudinal aspects.
5. The model should enable to consider possible driver errors, in addition to nor-
mative behaviour, in different traffic conditions. Error-generation mechanism
needs to be included in the model.
6. The overall model and simulation should be developed in such a configuration
to facilitate the evolution towards a 'real-time' simulation approach of the DVE
system.

The first two requirements concentrate on the specific objective of generating a

simulation tool for design and safety assessment, which is able to predict driver be-
haviour and interactions with the vehicle and environment in dynamic conditions.
The next three requirements focus on the model of the driver and, in particular,
concentrate on the need to consider cognitive aspects of behaviour, in addition
to manifestations and performances, when modelling the integrated DVE system.
In addition, adaptive behaviour and emotional and attitudinal aspects are also
identified as necessary peculiarities that need to be included in the simulation
since they are essential components of behaviour. Modelling human error is also
essential for evaluating mishaps and inadequate performance.
 19. Simulation of Driver Performance 349

The last requirement is specific to the perspective application of the simulation.

Indeed, even if the primary objective lies in the ability to support designers in
predictive assessments, it is equally relevant to identify how to adapt a predictive
model and simulation of this nature to more direct implementation in vehicles for
real-time application.
According to these requirements, the simulation should not necessarily 'run'
faster than real time. However, the ability to perform predictive studies of DVE
for safety purposes demands that many simulations are performed with different
behavioural characteristics and traffic/vehicle conditions. A balance between these
two requirements is therefore needed. Moreover, the consideration for errors and
dynamic aspects enhance quite considerably the need to enable the performance of
several simulation runs, even for single traffic configurations. Therefore, the overall
simulation has to present the basic characteristic of being predictive, simple and
fast running, accounting for dynamic interactions, human errors and adaptive
behaviour.
Amongst the variety of paradigms that enable to describe the DVE interactions,
those described in Chapter 2 of this book cover the vast majority of architectures
that have been proposed and are sufficiently developed to represent driver's interac-
tion in modern automotive systems and traffic contexts. The specific requirements
of the simulation discussed here favour the selection of the paradigm described by
Carsten (2007), in this volume, as it offers the possibility to represent a simple DVE
interaction, based on a quite normative type of approach to describe 'normative'
and task oriented behaviour. At the same time it depicts the framework to consider
cognitive and adaptive behaviour by means of a limited number of parameters
and offers the possibility to account for human error on the basis of these same
parameters. The specificity of dynamic interaction is also permitted, even if these
aspects are not actually discussed at modelling level. Also the characteristics of
real time and fast running are not reviewed, even if they are very relevant features
when transforming a paradigm into a running tool, at simulation level.
The simulation that has been developed on the basis of this paradigm is called
'Simple Simulation ofDriver Performance' (SSDRIVE). The specific characteris-
tics of the SSDRIVE will now be briefly described by discussing primarily the way
to represent normative behaviour, followed by the description of the algorithm by
which to consider cognitive aspects, behavioural adaptation and errors, and finally
by considering the dynamic interaction of the SSDRIVE within the overall DVE
simulation.

19.2.2 Simulation Approach/or Normative Behaviour

19.2.2.1 Task Analysis
The most suitable model for representing driver behaviour during the performance
of normative activities, both at primary and secondary levels, is to apply a simple
Task Analysis (TA) approach. The detail of accuracy of the analysis and description
of the tasks that are performed by the driver defines also the granularity of the
simulation (Michon, 1985).
350 Cacciabue, Re, and Macchi

By carrying out a driving TA, the performance of a driver can be formalised

and structured in a sequence of goals and actions that are carried out during the
interaction with the vehicle and environment. Commonly used task analysis tech-
niques such as Hierarchical Task Analysis (HTA) or Time Line Analysis (TLA)
(Kirwan and Ainsworth, 1992) are best suited to represent tasks that are associated
with procedures or routines in industrial environments. Driving on the other hand
is not suitable to a description following the principles of a traditional task analysis,
but rather requires a method that can cope with the irregularity of the driving situ-
ation. Driving can be easily seen as a set of interconnected and dynamically linked
set of goals, where the top goal is getting to the destination, while several series of
lower or simpler tasks are associated with local traffic situations. From this a set of
subgoals, corresponding means can be derived. Consequently, instead of proposing
a hierarchical structure of goals and subgoals, the functional dependencies among
goals are analysed and described. From this basis, a goals-means structure can be
instantiated for any given situation and set of conditions. While a HTA tend to
generate a single, typical task, a goals-means analysis produces a description of a
set of possible tasks. Therefore, the approach selected for the SSDRIVE simulation
to describe 'normative' driver behaviour is based on the rationale that identifies
the Goals-Means Task Analysis (GMTA) approach (Hollnagel, 1993).
In order to represent the set of Tasks that are carried during driving, a certain
differentiation has been defined according to the complexity associated with the
actual activity described by a Tasks. The following three basic elements have been
identified:

Elementary Functions, which represent the basic activity that cannot be further
subdivided into simpler components.
Elementary Task, which is a task made of elementary functions only.
Complex Task, which is a task made of a combination of elementary tasks and
elementary functions.

The first step to perform is to define the correlations between what has been
called 'task' and individual 'Elementary Functions'. Table 19.1 contains a set of
Elementary Functions that can be accounted for in order to describe the basic
actions of a driver. As an example, the Elementary Function 'Receive Perceptual
Input' is associated with visual, aural or haptic perception.
Table 19.2 shows a sample of correlations that are defined between tasks and
elementary functions. Several tasks have been developed for implementing the
simulation, such as 'Attain higher speed', 'Attain lower speed', 'Stop vehicle',
'Reverse vehicle', 'Tum left', 'Tum right', 'Change lane', 'Pass Vehicle', 'Over-
take', 'Keep lateral safety margins', 'Keep longitudinal safety margins', 'Maintain
speed', 'Give way at intersection', 'Emergency manoeuvre', etc.

19.2.2.2 Dynamic Logical Simulation of Tasks

Each task is represented in the simulation as a 'frame' (Minsky, 1975), which is
associated with attributes or pre-condition, which in tum enable the frame to be
initiated, and post-conditions, which are applied for 'closing' the frame.
 19. Simulation of Driver Performance 351

TABLE 19.1. Sets of elementary functions.

ELEMENTARY FUNCTIONS
Receive Perceptual Input Perceive phone ringing
Perceive warnings
Perceive indicators

Elementary checking actions Check mirrors
(primary task - vehicle controls) Check speedometer
Check road signals
Check signalsfor direction
Scan road side

Elementary control actions Accelerate
(primary task - vehicle controls) Brake
Steer
Change gear
Set indicator (right/left)

Elementary control actions Adaptive Cruise Control(ACC):
(primary task - vehicle ADAS) J Headway
J Check distance to lead vehicle
J CheckACC is enabled
J Select speed

The tasks, orframes, are simulated implementing the associated procedure and
sequence of elementary tasks/functions by means of sequences of rules and actions
described through object oriented programming languages. Attributes (pre- and
post-conditions) are correlated according to rules and entity-attribute matrices.
The logical dynamic sequence of tasks is defined according to a hierarchy derived
from the satisfaction of the pre- and post-conditions, coupled with a model based
on decision making and definition of intentions. This model is crucial for the
entire dynamic evolution of the DVE interaction. A simple model of this nature is
described in next section and may be associated with the maximum allowed speed
and on specific road and traffic conditions.
Another fundamental characteristic of the SSDRIVE simulation is the consid-
eration for the permanent or automatic tasks. These tasks are identified by the
fact that they are permanently carried out during a DVE interaction and do not
require specific pre-conditions to be launched. These are stereotypes of what may
be called 'skill-based behaviour' in a very 'classical' modelling architecture based
on a skill-rule-knowledge type behaviour that has been the most known imple-
mentation of the information processing paradigm proposed in the early 1980s for
describing human behaviour (Rasmussen, 1983). Two permanent tasks are con-
sidered in the present simulation approach, namely 'keep lateral safety margins'
and 'keep longitudinal safety margins' .
The concept of permanent tasks is quite straightforward, as it is associated with
the fact that drivers 'normally' keep their vehicle within lane margins and at a
reasonable safety distance with preceding vehicles or obstacles, without specific
352 Cacciabue, Re, and Macchi

TABLE 19.2. Examples of correlations between tasks and elementary functions.

Task/elementary
Goal/task functions involved Pre-conditions
Attain higher speed 1. Accelerate 1. No ahead vehicle with lower
(complex task) 2. Check speedometer speed
Goal/Post-condition: Reach 3. Maintain speed 2. All Preconditions of Elementary
the selected speed Tasks

Attain lower speed 1. Brake 1. Check all Preconditions of

(complex task) 2. (Change gear) Elementary Tasks
Goal/Post-condition: Reach 3. Check speedometer 2. Ahead vehicle with lower speed
the selected speed 4. Maintain speed 3. Presence of give way signals
Stop vehicle (elementary task) 1. Brake 1. Presence of Red traffic light or
Goal/Post-condition: Vehicle 2. Change gear 2. Presence of an obstacle on the
speed equal to zero carriageway
3. Car incoming from the main
road
Keep longitudinal safety 1. Scan road forward Permanent task
margins (complex task) 2. Check speedometer
Goal/Post-condition: Maintain 3. (Attain lower
safe distance from leading speed -emergency
vehicle manoeuvre)
Maintain speed (elementary 1. (Accelerate/ Brake) 1. No ahead vehicle with lower
task) speed or obstacle on road
Goal/Post-condition: Have the 2. Nored/yellow traffic light
previous time-step speed 3. No speed limitations signals
4. No give way at intersection
Entries in Italics indicate elementary function; entries in bold indicate task; entries within parenthesis
indicate conditional activity (function/task).
Elementary task = task made of elementary functions only. Complex task = task made of a combination
of elementary tasks and elementary functions.

effort and cognitive demand. Consequently, these two tasks are permanently active
in the DVE loop and are always performed every time the driver simulation is
activated. This simulation requirement, associated with the fact that there is no
cognitive load in performing these' simple vehicle control' tasks, in normal condi-
tions' has generated the choice of a classical (optimal) control modelling approach
(Weir and Chao, 2007, this volume) for their simulation in the DVE loop.
In practice, every time the driver simulation is activated in the DVE loop, the
permanent tasks are performed first, unless a human error is underway. They
aim at keeping the vehicle under control with respect to longitudinal and lateral
coordinated of the driving environment (road and traffic) (Fig. 19.1). The other
tasks are then performed according to the model of decision making and intentions
described above.

19.2.3 Algorithms for Cognition, Behavioural Adaptation

and Errors
The basic assumption made for the development of the driver model is that the
driver is essentially performing a set of actions on the vehicle commands and
 19. Simulation of Driver Performance 353

Step 2 Step 3

I
I
I
I I
I
DVE DVE DVE I
I
I
Interaction Interaction Interaction
I
I
I
I
I I
I I

I i I I I
o 2 4 ....a time

FIGURE 19.1. Dynamic-Logical Task sequence.

controls that are known and, in many cases, familiar, according to experience. As
the driving process is very dynamic, these actions are continuously selected or
developed from the knowledge base of the driver. However, prior to this activ-
ity a process of information management and formulation of goals and tasks is
necessary.
Consequently, independently of the specific model selected for describing the
driver behaviour, the following four steps of cognitive and behavioural interactions
must be considered:
1. Perception of signals from vehicle and environment.;
2. Interpretation of information.
3. Formulation of goals and intentions and selection of tasks to be carried out.
4. Performance of actions on control panel and on vehicle commands.
This is a typical formulation of a model based on the paradigm called 'Information
Processing System' (IPS), which has been applied in almost all fields to account for
human interaction with technology at different levels of automation (Cacciabue,
2004). The important characteristic of this approach is that it allows considering the
behavioural as well as the cognitive aspects of human performance. This paradigm
is the reference model for the SSDRIVE simulation.
The way in which the various cognitive functions of the model are implemented
in the simulation is quite simple and straightforward and depends on two possible
types of modelling architecture: the simple and linear normative driver behaviour
or the complex and more realistic descriptive driver behaviour. These two types
of simulation are now described in some detail.
354 Cacciabue, Re, and Macchi

Interpretation of signs and signals.

y
N

Intention development => Evasive Manoeuvre (Active Task)

Intention development => identify Active Task:

1. Active task is continued before starting a new task.
2. Task are started only when and if all pre-conditions are satisfied.
3. Speed is kept at maximum allowed by road signals and traffic.

Actions carried out according to Active task ~

FIGURE 19.2. Flowchart of the simulation of normative driver behaviour.

19.2.3.1 Normative Driver Behaviour

In normal conditions, i.e., when the driver has a very low or zero level of impairment
and no behavioural adaptation occurs, the simulation covers what may be called
normative driver behaviour. This is a merely theoretical condition not reflected in
realistic behaviours. However, from the simulation point of view it is necessary
that these formal aspects of driver behaviour are represented and stored in the
overall simulation tool.
The following sequence of steps is performed (Fig. 19.2):

1. All signals and signs that are produced inside and outside the vehicle are per-
ceived.
2. Interpretation is conforming to the meaning associated with signs and
signals.
 19. Simulation of Driver Performance 355

3. Intentions are formulated in relation to the ongoing task and information per-
ceived and interpreted. A task is then either continued or newly started. This
task is called active task.
4. Actions are carried out according to the active task.

The key step in normative behaviour is the process of formulation of intentions.

In the simulation, a very simple approach has been selected based on costlbenefit
rule for prioritising tasks. The rule implies that the driver minimises cognitive
efforts and time for reaching indented location. Therefore the following principles
apply:
1. Assess whether permanent tasks require a change of task or function (condi-
tional activity).
a. In the case of conditional activity, then adapt to demands of permanent task.
b. Otherwise continue the process.
2. The ongoing task is terminated before starting a new task.
3. Tasks are started only when and if all pre-conditions are satisfied.
4. Speed is kept at maximum allowed by road signals and traffic.

19.2.3.2 Descriptive Driver Behaviour

In non-normal conditions, i.e., when a certain impairment and 'error' may occur,
or when behavioural adaptation becomes a relevant factor, then the simulation cov-
ers what is called descriptive driver behaviour. In this case, the sequence of steps
remains unchanged as far as the information processing is concerned. However, be-
fore entering the IPS loop, the parameters governing adaptation and error must be
evaluated. This may lead to modifying some of the cognitive functions, including
the performance of actions. The latter represents the manifestation of adaptive or er-
roneous behaviour. Task selection and performance is also affected by behavioural
adaptation. This type of simulation is much more realistic with respect to actual
driver behaviour. However, the level of complexity from the simulation viewpoint
is much higher and requires some degree of simplification and linearisation.
In essence, descriptive driver behaviour is identified by a number of parameters
that enable to modify and adapt the way in which the more cognitive functions are
carried out, namely the perception and interpretation of signs and signals and the
generation of intentions. Moreover, the parameters are essential for the definition
of the actions that are performed.
According to the selected model, the following five parameters can be identified
for considering descriptive and adaptive behaviour:
Attitudes/personality (ATT): static parameter associated with each driver.
Experience/competence (EXP): static parameter associated with each driver.
Task demand (TD): objective dynamic parameter resulting from DVE interaction.
Driver State (DS): subjective dynamic parameter resulting from DVE
interaction.
Situation Awareness/Alertness (SA): subjective dynamic parameter resulting
from DVE interaction.
 356 Cacciabue, Re, and Macchi

TABLE 19.3. Identified correlations between parameters and measurable or observable

variables or the DVE evolution.
Parameters Definition MeasurableVariables
Experience The accumulation of knowledge or skills that 1. No of kilometresper year
result from direct participationin the 2. Number of years with
driving activity. drivinglicense
Attitudes A complexmental state involving beliefs and 1. Speed
feelings and values and dispositions to act 2. Lane keeping
in certain ways. Sensation Seeking and 3. Overtaking
Locus of Control have been identifiedas 4. Headway
personalitybased predictorsof accident
involvement.
Task demand The demands of the process of achieving a 1. Traffic complexity
specific and measurablegoal using a 2. Weather
prescribedmethod.When TaskDemandis 3. Light
focused only on driving,then Task 4. Speed
Demand = DrivingDemand. 5. Drivingdirection
Driver state Driverphysical and mental ability to drive 1. Lane keeping;headway
(fatigue,sleepiness... ). A set of dynamic control
parametersrepresentingaspects of the 2. Durationof driving;
driverrelevantfor the human-machine time-on-task
interaction. 3. Weather; Road conditions
4. Traffic complexity
5. Speed
Situationawareness Perceptionof the elementsin the 1. Distraction
environment within a volumeof time and 2. DriverState
space, the comprehension of their meaning 3. TaskDemand
and the projectionof their status in the
near future.

The discussion on the reasons for selecting these parameters accounting for de-
scriptive driver behaviour goes beyond the scope of this paper and can be found
in Carsten (2007, this volume). The definitions that are associated with each pa-
rameter are shown in Table 19.3. However, it is primarily noticeable that the two
static parameters of Attitude/personality and Experience/competence enable the
definition of some initial characteristics of behaviour that affect all decision mak-
ing processes and performances throughout the simulation. On the other side, the
dynamic parameters enable to account for the variability of conditions and envi-
ronmental changes that are encountered during a typical journey. In this sense,
both types of parameters are very important and may playa crucial role in the sim-
ulation of DVE interactions. Naturally, this implies that the model and associated
simulation are sufficiently powerful to incorporate all these effects.
It is important to note here that the parameter Task Demand is in most cases
equivalent to Driving Demand. However, as the objective of the correlations be-
tween parameters and behaviour covers all potential aspects affecting driver per-
formance, it is preferable to retain in this parameter also the factors associated
 19. Simulation of Driver Performance 357

with secondary task performance, including social and personal demands derived
from the overall vehicle context.
The set of variables that govern the dynamic driver model is directly correlated
with the ability of the model to account for the above functions at different levels
of depth. The crucial open issues that remain to be resolved with respect to the
DVE parameters are of three types:

1. Correlation ofparameters with respect to measurable variables. This involves

two aspects:
Which are the actual measurable variables that affect the parameters.
How these variables influence the change of value of the parameters.
2. Intentions/goals. The dynamic process of decision making and interaction with
the Vehicle-Environment is regulated by the development of intentions, such
as 'intention to overtake a vehicle in front', 'increase speed' , etc.
3. Correlation ofparameters with respect to driver behaviour. How the parameters
combine in affecting driver behaviour and/or performance.

19.2.3.3 Parameters and Measurable Variables

The next issue that needs to be resolved in order to progress with the development
of a simulation is the identification of the way in which parameters affect driver
behaviour. This issue, as stated above, involves two aspects: the definition of which
measurable variables are associated to each parameter and how such variables
affect parameters.
Since we are here interested in the implementation of the SSDRIVE from the
simulation perspective, the theoretical discussion associated with the first part of
the problem is of less interest. Consequently, we will not discuss the reasons and
scientific findings about the selection of the variables that are associated to each
parameter. The reader should refer to the specific literature in this area and in par-
ticular to Chapter 2 for a more specific treatment of this subject. Table 19.3 contains
a set of variables that can be measured, or evaluated in the case of a simulation,
which enable the estimation of the dynamic values associated to each parameter.
The problem of interest in the case of the development of a simulation is the
numerical or analytical expressions that are applied for the evaluation of the pa-
rameters on the basis of the calculated values from the dynamic simulations of the
vehicle, of the environment, and also from the driver itself.
The initial values associated to the Attitude/personality and Experi-
ence/competence do not change over time, as they are static parameters and are
simply input data for the SSDRIVE simulation. Static parameters influence, to-
gether with the dynamic values associated with the other three parameters, driver
decision making and behavioural aspects, as it will be discussed in next section.
Focusing on the dynamic parameters and on possible forms of correlations
between them and measurable variables, the most generic representation that can
be formulated, according to the quantities identified in Table 19.3, are as follows:
358 Cacciabue, Re, and Macchi

Task demand: TD = f (Traffic Complexity, Weather, etc.)

Driver state: DS = g (Lane Keeping, Duration ofDrive, Weather, etc.)
Situation awareness: SA = l (Driver State, Distraction, Task Demand, etc.)

The forms that these functions may take are numerical or logical expressions.
In the SSDRIVE approach, the choice of utilising fuzzy descriptions has been
made. Consequently, during the simulation, each (independent) variable of inter-
est is evaluated by a numerical correlation and then associated to a fuzzy value.
Subsequently, the estimation of the (fuzzy) value of each parameter requires the
application of typical fuzzy rules that combine the (fuzzy) values of the associated
independent variables.
In practice, the five parameters that govern the SSDRIVE are fixed. A set of
default correlations between parameters and measurable independent variables is
available. These require in input only the limits for each independent measurable
variable to structure the fuzzy correlations. The fuzzy rules that define the values
of the parameters are fixed, while screening on the fuzzy correlations associated
to each independent variable is possible via input data to the simulation tool.
On the other side, if the user of the simulations intends to study and apply
new functions and rules, then the input requires the definition of new independent
variables, to be selected amongst those calculated by the three modules of the DVE
simulation, and the formulation and programming of new correlations that replace
the default ones. This is a very flexible aspect of the simulation tool that is based
on a 'modular' structure. However, this requires programming ability and detailed
knowledge of the simulation software.
The correlations adopted as default for each dynamic parameter will now be
briefly discussed.

19.2.3.3.1 Task Demand

The default fuzzy correlations and the rule for the evaluation of Task Demand are
(Fig. 19.3) as follows:

1. Task Demand (TD) is assumed to be a fuzzy function with values High, Accept-
able, and Low, and is correlated only to visible level of traffic and environment
(Visibility), and to the number of vehicles present on the road (Complexity of
traffic, CoT):
TD = f (Visibility CoT)
2. Visibility is measured in terms of distance and is considered
Good, when the driver can see at a distance of more than 100 metres;
Acceptable, for distances of clear vision between 70 and 100 metres;
Bad, for less than 70 metres of clear vision.
3. Complexity of traffic (CoT) is measured in terms of vehicles per kilometre on
the road:
Good, when the density of vehicles is less than 100 per kilometre;
 19. Simulation of Driver Performance 359

Task Demand (TD) Variables (v) Fuzzy Values

Visibility (Vis) Good Acceptable Bad
Complexity of traffic (Co T) I Good Acceptable Bad

Vis CoT

G G

A A

B B
70 100 m 100 250 No. of
vehic1es/Km

High if mineVisib, CoT) = Bad (at least one v = Bad)

TD Acceptable if Bad < mineVisib, CoT) < Good (both v > Bad, & at least one v < Good)
{
Low if mineVisib, CoT) = Good (both v = Good)

FIGURE 19.3. Example of fuzzy correlations and fuzzy rule between Task Demand versus
Complexity of traffic and Weather Conditions (Visibility).

Acceptable, when the density of vehicles is between 100 and 250 per kilome-
tre; and
Bad, when the density of vehicles is greater than 250 per kilometre.
4. The fuzzy rule applied to calculate the fuzzy value of task demand is
TD == High if min (Visibility, CoT) == Bad;
TD == Medium if min (Visibility, CoT) == Acceptable;
TD == Low if min (Visibility, CoT) == Good.
This means that the less favourable value is associated with TD between the
two fuzzy values of Visibility and Complexityof traffic.

19.2.3.3.2 Driver State

The default fuzzy correlations and rule for the evaluation of Driver State are shown
in Fig. 19.4.

1. DriverState (DS) is assumed to be a fuzzy function with values Bad,Acceptable,

and Good, and is correlated simply to the time dedicated to driving (Duration of
Driving, DoD), and to the Speed change (I~Speedl) and Steering (I~cpl), over
a period of driving of 10 minutes:
DS ==f(DoD, I~Speedl, I~cpl)
2. Duration ofDriving (DoD) is measured in terms of overall driving time and is
considered
Low, when the driver has been driving for less than 60 minutes;
Acceptable, for a driving time between 60 and 180 minutes;
High, for more than 180 minutes of driving time.
360 Cacciabue, Re, and Macchi

Driver State (DS) Variables (v) Fuzzy Values

Duration of Drive (DoD) Low - L Acceptable - A High -H
IL1Speedl l O' Low - L Medium - A High -H
'L1q> '10' Low - L Medium - A High -H

DoD- IL1Speed I

L i
H I
H
I
,
1

A A I A
I i
I
H ----~-----
I
L
I
I
1---------1----------- L I
---------1-----_----
i
60 180 min 10 30 Km/h deg

Bad } High
DS Acceptable if Min [ Max( I L1Speed I , I L1q> I), DoD} Acceptable
{ Good Low

FIGURE 19.4. Example of fuzzy correlations and fuzzy rule between Driver State versus
Duration ofDrive, Speed change and Steering.

3. Variation of steering (I ~cp I) is associated to the road geometry and not to the
yaw disturbance. Variation of speed (I~Speedl) is associated to the traffic con-
ditions and road type/geometry. Therefore, it is assumed that in the case of
roads with constant geometry (straight roads or small bends and curves, e.g.,
highways) very little change of steering and speed occurs. These are causes of
possible drowsiness in drivers. Changing either of them, over a period of 10
min, is considered sufficient to retain an acceptable level of vigilance. They
are therefore combined in defining a fuzzy measure of Driver State. The fuzzy
correlations for the variation of speed and steering (for a period of driving ~t =
10 minutes) are as follows:
Low, when I~Speedl < 10 km/h or I~cpl < 5.
Medium/Acceptable, when 10:s I~Speed I :s 30 km/h or 5 :s I~cpl :s 30.
High, when I~Speedl > 30 kmlh or I~cpl > 30.
4. The fuzzy rule applied to calculate the fuzzy value of Driver State is as follows:
DS = Bad if = Low
DS = Acceptable if min [max(I~Speedl, I~cpl), DoD] = Acceptable
DS = Good if = High
This means that the less favourable value is associated with DS between the
fuzzy values of Duration ofDriving and the combination of Speed change and
Steering.
 19. Simulation of Driver Performance 361

Situation Awareness Variables (v) Fuzzy Values

Distraction (Dis) Low Medium High
Task Demand (TD) Low Acceptable High
Driver State (DS) Good Acceptable Bad

Dis (Vj)

L L
50 100 200 250 Vj 3 5 v2
(no. of veh./km) no. of active
indicators & IVIS
Dis = f[no. ofveh./km (vI)' no. of active indicators & IVIS (v2)]

High if max (v V2) = High (least one v = High)

Dis Medium if Low < max (vb V2) < High (both v < High, & at least one v > Low)
{ if max (vb V2) = Low
Low (when both v = Low)

SA -j.at SA = h(DS)
Bad if TD=Low Bad if DS=Bad
SA Good if TD = Medium SA Acceptable if DS = Acceptable
{ {
Bad if TD=High Good if DS = Good

SA = h(Dis)
Bad if Dis = High
SA Acceptable if Dis = Medium
{
Good if Dis = Low

SA =f(Dis, TD, DS)

Bad if min [Jj(TD), h(DS), h(Dis)} = Bad
SA Acceptable if min[Jj(TD), f2(DS), f3(Dis)} = Acceptable
{
Good if min[Jj(TD),h(DS),h(Dis)} = Good

FIGURE 19.5. Example of fuzzy correlations and fuzzy rules between Situation Awareness
versus Driver State, Task Demand and Distraction.

19.2.3.3.3 Situation Awareness

Finally, default fuzzy correlations and rules for the evaluation of Situation Aware-
ness are shown in Fig. 19.5.

1. Situation Awareness (SA) is assumed to be a fuzzy function with values Bad,

Acceptable, and Good and is correlated to Task Demand (TD), Driver State
(DS), and Distraction (Dis):
362 Cacciabue, Re, and Macchi

SA =!(TD, DS, Dis)

2. Task Demand (TD) and Driver State (DS) are measured according to the two
fuzzy rules and functions described above.
3. Distraction (Dis) is a fuzzy rule, associated to two fuzzy functions, i.e., number
of vehicles per kilometre on the road (VI) and number of active indicators and
IVIS (V2):
Disv1: High when VI < 50 or VI > 250
Medium, when 50 ~ VI ~ 100 and 200 ~ VI ~ 250
Low, when 100 < VI < 200
Disv2: High when V2 > 5
Medium, when 3 ~ V2 ~ 5
Low, when 3 < V2
Then the fuzzy rules imposes that the less favourable value is associated
with distraction between the two fuzzy values of number of vehicles per
kilometre on the road (VI) and number of active indicators and IVIS (V2):
Dis = Max o; Dis v2)
4. The fuzzy rules that combine SA with each variable TD, DS and Dis are as
follows:
!1 (TD) = SA (TD)
SA (TD) = Bad if TD=Low
SA (TD) = Good if TD=Medium
SA (TD) = Bad if TD = High
!2 (DS) = SA (DS)
SA (DS) = Bad if DS = Bad
SA (DS) = Acceptable if DS = Acceptable
SA (DS) = Good if DS= Good.
!3 (Dis) = SA (Dis)
SA (Dis) = Bad if Dis = High
SA (Dis) = Acceptable if Dis = Medium
SA (Dis) = Good if Dis = Low
5. Finally, SA is evaluated according to the following fuzzy rule:
SA = Bad if min f!I(TD),!2(DS),!3(Dis)] = Bad
SA = Acceptable if min f!I(TD),!2(DS),!3(Dis)] = Acceptable
SA = Good if min f!I(TD),!2(DS),!3(Dis)] = Good
This means that the less favourable value between the three fuzzy values of
Task Demand, Driver State, and Distraction is associated to SA.

The complete set of fuzzy correlations and fuzzy rules that govern the SS-
DRIVE simulation is quite complex and requires a considerable effort of data
input and data definition. This is not surprising as the actual combinations of vari-
ables, parameters and effects are distributed and cannot be oversimplified with
 19. Simulation of Driver Performance 363

the risk of undermining the overall effort of modelling carried out at theoretical
level.
The actual formulation of the parameters that govern the model of the driver is
obviously critical for the overall DVE modelling. For this reason, the implementa-
tion of the correlations linking all static and dynamic parameters, i.e., EXp, AIT,
DS, TD, and SA, and measurable variables is kept wide open in the simulation
approach of SSDRIVE. This means that the users have two main options:

1. To utilise the set of default correlations, i.e., those that are described in this
paper, and adapt simply their boundary conditions and limits by input data.
This alternative is obviously easy and simple.
2. To apply a set of specific correlations, by developing appropriate programming
routines that will then be interfaced with the main SSDRIVE simulation. This
makes the input process more complicated and time consuming. However, in
this way, the flexibility of the overall simulation is widely improved and makes
possible to test different formulations and combinations of (fuzzy) functions
and rules that combine parameters and variables.

19.2.3.4 Intentions, Decision Making and Human Error

The overall simulation is governed by the dynamic (time dependent) unfolding of
intentions and decisions of the driver model. This is another crucial aspect of the
model, from the theoretical viewpoint, as it impacts on the validity of the overall
approach and relies on the enormous amount of research in this area, in association
with humans attitudes, characteristics, personality, individual and social aspects,
etc. Many of these issues are discussed in various papers of this book. The literature
on this matter dates back several decades and is rich of many different theoretical
stands and formulations.
From a simulation perspective, the choice that has been made in the development
of SSDRIVE is to keep maximum flexibility, as it has been done in the case of the
definition of the parameters and measurable variables. The user will retain the
possibility either to apply a set of default options or to separately develop parts of
the SSDRIVE so as to modify the default settings. As in the case of parameters
and measurable variables:

1. the default settings are kept as simple as possible in order to make the develop-
ment of standard cases relatively easy and fast and
2. the development and implementation in the simulation of new algorithms will
require a more complex definition of the input and deeper knowledge of the
simulation architecture and programming languages.

The set of defaults conditions existing in the simulation are now briefly dis-
cussed. These are simple correlations that govern primarily development of inten-
tions and decision making, and error generation.
364 Cacciabue, Re, and Macchi

19.2.3.4.1 Intentions and Decision Making

The generation of intentions is simulated by means of a very simple correlation
that enables the dynamic simulation of tasks and elementary functions and new
intentions and goals, as the overall DVE interaction progresses and new traffic
conditions are met.
At the default level, the correlation that has been chosen is very simple and it is
based on the driver assessment of the cost-benefit of the minimum time to reach
the objective, i.e., going from the starting point of the journey to its destination in
the minimum allowable time. In other words, the vehicle is driven at the highest
'intended' speed, depending on driver characteristics, traffic and maximum allowed
speed. The static parameters of experience and attitude will affect the highest
intended speed in a very simple manner. The decision making process based on
the two parameters EXP and ATT would give a constant and thus static value of
highest intended speed with respect to the maximum allowed speed.
In order to introduce a dynamic contribution to this decision making process,
two other parameters are introduced in the correlation, namely, TaskDemand (TD)
and Road condition/geometry. The overall correlation that is set as default for the
intended speed (Speedintended) is

Speedintended == {I + O.2[f(EXP, Am + g(TD)]

+ O.21/J(Roadcond.)}* Speedmax-allowed.
The following data and functions apply:
A1Thas three possible fuzzy input values EXP has also three possible input data
1. ATT == 0, for Low-Risk-Taker 1. EXP == 0, for Low-Experience
2. ATT == 1, for Moderate-Risk-Taker 2. EXP == 1, for Moderate-Experience
3. ATT == 2, for High-Risk-Taker 3. EXP == 2, for High-Experience
f(EXP, A1T) == (ATT - EXP + 1) -1 ~ f ~ 3
and

ifTD == High --+ g (TD) == -1

ifTD == Medium --+ g (TD) == 0
ifTD == Low --+ g (TD) == 1

and
ifRoadcond == Bad --+ 1/J(Roadcond.) == - 1
if Roadcond == Acceptable --+ 1/J(Roadcond.) == 0
if Roadcond == Good --+ 1/J(Roadcond.J == 1.
In this way, the intended speed may vary between the two following maximum
and minimum values:
Max Speedintended == 2 * Speedmax-allowed
and

Min Speedintended == 0.4 * Speedmax-allowed

 19. Simulation of Driver Performance 365

The consequent dynamic unfolding of tasks, such as overtaking of slower vehicles

and avoiding obstacles, depends on this simple process of decision making asso-
ciated with the intention of the driver to select and maintain the intended speed
(Speedintended), as the limits change along the road. As discussed earlier, also in this
case, the possibility for the user to utilise different and more complex correlations
is granted by the specific type of simulation.

19.2.3.4.2 Error Generation

The last open issue concerns the correlation of the five basic parameters with driver
behaviour and performance with respect to error generation. As in the previous
cases, the user will have the possibility to apply different and complex correlations
by exploiting the flexibility of the simulation tool.
At the default level, very simple assumptions are made, associated with the
dynamic evolution of the DVE interaction. The mechanism that has been devised
to describe driver error and behaviour in relation to the basic parameters is called
Model of Basic Indicators of Driver Operational Navigation (BIDON). To provide a
first attempt of evaluation of dynamic change ofDVE conditions, it is assumed that
the variables affecting the driver behaviour, i.e., the subjective dynamic parameters
DS and SA and the objective dynamic parameter TD, are represented by 'contain-
ers' with thresholds/levels, which change from a driver to another and enable to
define the overall state/performance-ability of the driver, as the DVE interaction
evolves. The static parameters Experience/competence and Attitudes/personality
are evaluated at the beginning of a simulation and remain constant. They also
contribute to the initial 'filling' of the 'containers' and can therefore be associated
to the starting levels of Situation Awareness (SAo), Driver State (DSo) and Task
Demand (TD o).
Every time a relevant event happens, the level of both subjective and objective
conditions will change, affecting the efficiency and performance of the driver
behaviour. This will contribute to the dynamic process of progressive 'filling' or
'draining' of the 'containers' of the BIDON model.
The error-generation mechanism that is implemented in the SSDRIVE is as
simple as the other correlations that have been discussed above. The same
rule of flexibility applies, whereby the user of the simulation tool is able to
make use of more complex error making functions by means of more compli-
cated formulations from input data setting and definition of specific programs
that may be interfaced with the main simulation. At default level, the error-
generation process is essentially associated to a single parameter called Driver
Impairment Level (DIL). The DIL depends essentially on the static and dynamic
parameters.
The values of Experience/competence, Attitudes/personality, Situation Aware-
ness (SAo), Driver State (DSo) and TaskDemand (TDo) are user input. The DIL or
error-generation mechanism is associated to the following correlations:

DIL t = !CSA t , DS t , TD t , EXP, AIT) 0::::; DIL::::; 1.

366 Cacciabue, Re, and Macchi

The following logics and fuzzy correlations are implemented as default:

1. An error occurs when DIL == 1.

2. DIL == 1, if
all three dynamic parameters reach their most negative conditions at the same
time, i.e., SAt == Bad, TDt == High, DSt == Bad; or
anyone of the three dynamic parameters reaches and remains at its most
negative condition for at least N consecutive minutes (~tDIL). ~tDIL depends
on the age of the driver and on the parameters EXP and AIT. The following
fuzzy rule applies:

~tDIL == N + !(EXP, Am + g(AGE),

where N == 4 (default value)
-1 2 if min(EXP, Am == High
! { 0 if Moderate < min(EXP, Am < High
if min(EXP, Am == Low
and
- I if AGE> 50
g 0.5 if 35 < AGE < 50
{
a if AGE ~ 35
In this way, the maximum and minimum time intervals for error making when
one of the critical parameters is at its most negative condition are

Max ~tDIL == 5 minutes

Min ~tDIL == 1 minute
When an error occurs, then the types and modes of errors are associated with the
ongoing activity being carried out by the driver at the time of the error. The types
and modes of errors are also defined through the input data system. However, the
simulation is able to combine different errors, situations and solutions.
Moreover, the whole variety of potential errors (types and modes) that may
occur when the DIL == 1 needs to be studied in order to enable the evaluation of
the largest variety of DVE dynamic interactions. This concept is described in the
example shown in Fig. 19.6, where the first error occurs in time-step 6 (SAt = 6 ==
Bad, DSt= 6 == Bad and TDt= 6 == High). In this case, two possible types/modes of
errors have been identified by input data and consequently, three sequences are
generated: one with no error; one with the first type/mode of error; and the third
one with the second type/mode of error. This generates a very complex spectrum
of possible sequences and requires a very accurate analysis tool for screening, the
selection of relevant results, and outcome of the simulation.

19.2.4 Simulation of Control Actions

The overall DVE and SSDRIVE simulation processes are shown in Fig. 19.7 and
Fig. 19.8. In particular, at each time interval, the driver model receives the 'real'
 19. Simulation of Driver Performance 367

..- ..- ..
... - '"

I I I I
o 2 4 6 8

I
CD"~~:~_ " T"
A 8
DS D" "
river Pa rameters at time t4

.. TO

~
C
DS

F IGURE 19.6. Error mechanism and dynamic sequences generation.

vehicle and environment conditions, i.e., the dynamic variables calculated at pre-
vious time steps, and evaluates the new actions that are carried out on the vehicle
controls (Fig. 19.7, upper part). These are essentially the settings of indicators,
ADAS, IVIS etc., and the basic vehicle control action s of steering and acceler-
atinglbraking in order to obtain to the desired position and speed. The cognitive
processes and behavioural adaptation that lead to decisions have been discussed
in the previou s sections. In the following, the implementation of control actions is
discussed in some detail s.

19.2.4.1 Normal Driving

In order to implement the performance of tasks and actions on the vehicle control s,
few robust driver control theory models have been chosen. The first model is the
Crossover Model (McRuer and Krendel , 1974; see also Chapter 17 of this volume),
which is described in term s of steering control and acceleration pedal. The second
model is the Shaft Model of driver 's control strategy (Kondo, 1953). A control
theory based approach is an efficient and low cost way to perform simulations of
DVE interactions. In particular, it allows to account for perturb ations associated
with behavioural aspects such as noise on steering control, delay in traje ctory
planning, and delay and loss of recognition of target information .
 368 Cacciabue, Re, and Macchi

Ifeath er conditions
Road conditions
jDR IVE R Made/l

Traffic
.....

C01ztro/n1s:~tD.4S.J'f1'r'd Control M SiADAS. dn w

Perception
----------------- Brake Pedal Fo~
".inm.J Actions
Interpretation Accel. Pedal F_
Posifdfl ,,m Optimal Control
Planning (Intentions) Steerin g Angle

t f eedback
I

----------------,
r- ! I
I
F....., IJriw"V.Iud- Modd

00 .. . .... : . . . ... : . : . . . . .. ./. + - - - ' -----l

I
U/. tJ ' +~ l4(I)
,
j FrotIf
l Driwr
i EIW;n;HtIrW,1I
70 fi: i .
i .IIod. t
~ 60 ------t-------t------ :------i--------- ----------
50 --.- ---- , -- --- - --- ,-- --- -----. - -- -------

40 _ . ~ . __ A ~ _ _- : :

10 20 40 50

FIGURE 19.7. Control theory model of the SSDRIVE simulation and desired speed profile
during transition from 50 to 80 km/h with a Tstyle of 20 s.

Environment

r.
~

Env ironmental
Parameters
Obstacles I
Vehicle speed Weather condrtions
Vehicle position Traffic and road conditions
...... ~Ir
......
Simulation
Dr iver
I ~
Manager
!Dr ive r' s Parameters Task analysis &

,,
Driver's Actions
Next LH OIL
I 11\
Lane position (steering angIe)
Vehicl~ speed
Speed profile (brake/accelsrator)
Vehicle posrtion
Control setlings (ADAS, IVIS,
ADAS interface
indicators, ......)
IVIS interface ,Ir
...... ....
Vehicle Dynamic

FIGURE 19.8. Overall dynamic DVE interaction process.

 19. Simulation of Driver Performance 369

The detailed description of the control theory model applied within the SS-
DRIVE simulation goes beyond the scope of the present chapter and can be found
elsewhere (Re and Carusi, 2005). In brief, the simplified driver behaviour rules
applied in the simulation are associated to the desired lane position (lpd) and de-
sired speed profile (Ud) (Fig. 19.7). These quantities are evaluated according to
the following rules:

Desired Lane Position (lpd): the car maintains the centre of the lane.
Desired Speed Profile (U d ) : while the rule for desired lane position is regulated
also in terms of driving good practice, to define the ideal speed profile is not
equally simple. Taken into account the rule of minimising travel time according
to driving at the maximum allowed speed (which means that after the transition
Ud is almost constant in a highway), the desired driver speed profile is obtained
by means of a smooth profile. This allows to increase or decrease speed after the
decision of attaining higher or lower speeds taken by the more cognitive part of
the simulation.

During the transition from a speed limit to the successive speed limit, Ud is
modelled as UTransition:

o-:: (t)
_= ~ (1 +
V
sin(wl - n
2
/2)) '

where w = n /Tstyle and 1 = 0 -:- Tstyl e. The outcome of the optimal control process
is then transformed into steering angle and actions on accelerator/brakes. The
optimal control model does not affect the positions of controls and indicators of
ADAS/IVIS and other settings, as these are evaluated by the cognitive part of the
simulation.
The overall dynamic interaction within the DVE simulation of the SSDRIVE is
finally summarised in Fig. 19.8. In particular, at each time step of the simulation,
the driver model calculates the implementation of tasks and actions and generates
profiles of steering angle and actions on the break or the accelerator, in order to
accommodate for the desired lane position and speed profile.
The Simulation Manager is the governing module of the simulation, which
receives input from all three main DVE components and evaluates the overall
error-generation mechanism by means of the DIL and keeps account of the syn-
chronisation of the simulation. The effect of ADAS and IVIS, if contained by the
vehicle simulation, is also fed to the simulation manager.

19.2.4.2 Error in Control Actions

In the case of the SSDRIVE, moving from a control theory model extracted from
assessed real data and designed in the continuous time domain to the implemen-
tation in discrete time domain without real mechanical constraints required some
mathematical manipulations. In this way, it was possible to ensure the consis-
tency between the cross-over model evaluated on test and the simulation. First
of all the formulas were all merged from the Laplace S Transform to the time
370 Cacciabue, Re, and Macchi

Ud(S) + U(s)
----...-j.. . ( X ) - - - - + 1

Ipd(S) +
--+0

lp(s! I.. _
FIGURE 19.9. Control model in terms of the equivalent inner open loop Laplace transfer
function of heading angle control.

domain. Following the Astrom-Wittenmark book, an anti-windup regulator has

been added. This filter was proven necessary in order to avoid unrealistic output
in certain condition, e.g., discontinuity due to U turns.
The actual formulation of the Open Loop Transfer Function for Speed Regula-
tion, Tspeed(t), for Lane Position and Speed Control, TzpJpeed(t), for Lane Position
Regulation with constant speed, Tzp_const Jpeed(t), and the Closed Loop Transfer
Function for Heading Angle Regulation, Tha(t), are shown in Fig. 19.9.
In order to simulate inadequate driver behaviour, the available variables are
associated with the output of the driver model, namely actions on the vehicle
controls, brake/accelerator pedals and steering angle (Fig. 19.7). The error types
are defined at modelling level. While the actual modes, i.e., the actual magnitude
of the errors, are defined in terms of amount of inappropriate steering, speed
selection (brake/accelerator), IVIS/ADAS setting, etc., are defined at input level
of the simulation. These quantities are implemented in the simulation through
the control model by either inhibiting the appropriate feedback loop or by setting
erroneous desired speed and vehicle positions.
The following are typical examples of errors that can be simulated and are
defined in input:

Incorrect settings of ADAS/IVIS.

Improper acceleration/braking (b.Ud ) .
Insufficient or excessive steering (b.lpd).

Discrete values of these quantities must be assigned in input in order to allow the
simulation of limited sets of sequences of events.
An example of improper steering, due to driver distraction or other causes, is
shown in Fig. 19.10, where a driver is simulated and will not correct the heading
angle by means of the steer controller and will not decrease the speed for a certain
time (Tdistraction). This type of approach has also been attempted by Salvucci (2001).
However, the SSDRIVE approach offers a wider variety of inadequate actions and
causes as well as permitting the application of novel and diversified correlations
of human error generation.
 19. Simulation of Driver Performance 371

~.+ iuo
~
U(s)

~ lp.(s)

7S
O
Constant control
= value applied by
the Driver
x = Ignoredinput

FIGURE 19.10. Simulation algorithms of inappropriate steering and speed control.

19.3 Sample Cases of Predictive DVE Interactions

The two sample cases discussed hereafter have the only objective to demonstrate
the functionality of the simulation, rather than discussing detailed applications of
predictions of DVE interactions according to well-defined scenarios and dynamic
situations. The reason for such type of approach is dual: firstly, it would be too long
to discuss in detail an application that attempts to predict DVE interactions, as the
overall scenarios setting and DVE correlations would require an extended discus-
sion and a accurate scientific justification. Secondly, the results of the simulation
runs should also be analysed for their implications on the design and associated
safety issues.
However, these issues go beyond the objectives of this paper, which is focused
at demonstrating the availability of a DVE simulation tool, centred on the driver
behaviour modelling, and at showing its potentialities in terms of diversity of
behavioural adaptation, error generation and parametrical correlations. For these
reasons, the following two sample cases discuss rather simple situations and highly
theoretical situations that stress the computational power of the simulation rather
than representing realistic behaviours of drivers in similar situations.

19.3.1 Case 1
The first sample case discussed here is based on a simple path considering a series
of round turns and multiple lanes roundabouts trajectories. The lateral and angular
deviations have been evaluated. The path is composed of two turns, one anticlock-
wise and one clockwise, and several times lane change on a three-lanes roundabout.
No human error or inadequate behaviour is considered. The driver simulation fol-
lows the normative behaviour. The desired and actual paths calculated by the driver
lane control model are illustrated in Fig. 19.11. It is clear that the desired path and
the actual position of the vehicle are almost completely overlapping, even in the
presence of a rather complex trajectory.
 ...,
-.J
N

n
~
(")
S'

. ': ~
25 - - Desired Path
- - Actual Path
~
20 ~
::I
15% 0..
15 "
z~
10 (")
10%
e:
5

g 0 5%

-5
0%
-10

-15
O.606Oll
-20
Steeri>J Angle
-25 Adjustment I'ad]
L
0 10 20 :J] 40 50 60 70
[m) O.om

FIGURE 19.11. Study Case I : Vehicle path analysis and centre lane deviation against steering angle adjustment.
 19. Simulation of Driver Performance 373

As the speed profile was chosen constant, the effect of the adjustments of steering
angle versus lateral deviation from the desired centre of lane could be evaluated.
Negative values of steering angle correspond to steers in the right direction. As
shown in Fig. 19.11, there are no particular trends in terms of steering angle adjust-
ment and centre of lane lateral deviation. The histogram contains the distribution
of the samples. The peak (20%) is around position (0.0; 0.0) and represents the
samples coming from straight sections of the path. Most of the samples (>90%)
are concentrated in the angular adjustment interval -0.01 to +0.01 radiant and
more than 95% of samples have a centre of lane lateral deviation less than 0.29 m.

19.3.2 Case 2
The second sample case discussed here is the performance of the same paths
simulated in the first sample case, where the difference is introduced in terms
of a loss of control. We will not discuss here the reasons for such human error
since the goal of the case study is to demonstrate the power of the simulation
to represent the consequences of erroneous behaviour. The simulation of loss of
control is performed by modelling the error as a freezing of steering control. As a
consequence, no further adjustment of the steering angle occurs in order to follow
the road centre lane.
As in the previous case, the desired and actual paths calculated by the driver
lane control model are reported in Fig. 19.12. It is clear that at a certain instant,
the desired path and the actual position of the vehicle separate completely and,
unless a recovery action takes place, the vehicle is actually driven off the road. In

-2 r---,---_r--_~___,--::--r ____,_-____,_-____r-_r__------___,
- - Desired Path
-4 - - Actual Path
-+- Distraction Path
- - Lane Boundaries
-6

-8
I
-10

-12

-14

-16

-18 L . . . - _ L . - - - - - I I . . . . . - - - - - I _ - - - - I _ - - - - J . _ - - - - ' - _ - - - - ' - _ - - . L . _ - - - L . _ - - - U

ffi ~ ~ ffi ~ 00 ~ M ~ ffi m
[m]

FIGURE 19.12. Study Case 2: Vehicle path analysis in the case of erroneous steering.
374 Cacciabue, Re, and Macchi

this case, no recovery action was simulated and the overall DVE interaction was
stopped at the time of expected vehicle loss of control.

19.4 Conclusions

In this paper, we have described the SSDRIVE. While the governing model and the
basic theoretical considerations are derived from a specific paradigm described in
referto thepaper (Carsten, 2007), in this paper we have discussed the set of default
correlations and algorithms that are implemented in the simulation. The overall
architecture of the SSDRIVE simulation has been conceived with the objective
of maximum flexibly from the user point of view. In other words, the simulation
is based on a modular structure and the user is able to include in the simulation
different models and modules describing specific driver support and information
systems, in accordance to specific needs or design options.
Following a set of preliminary tests, the model can be considered stable and
reliable. The development of the model and simulation will continue in terms of
deviation analysis during control delay and control freezing conditions. In partic-
ular, the analysis of a variety of IVIS and ADAS interventions will be gradually
included in the simulation, following the same principle of maximum flexibly for
the user. In this way, the core features of the SSDRIVE simulation will remain
unchanged, in terms of basic modelling architecture and fundamental principles,
such as the consideration for task analysis process, normative versus descriptive
driver behaviour, number of parameters that govern driver behaviour and basic
DVE interactions. At the same time, thanks to the modularity of the simulation ap-
proach adopted, the users of SSDRIVE will be able to test specific configurations,
typical proprietary models of ADAS and IVIS, as well as particular algorithms for
human errors and correlations between parameters and environmental variables.
In the present configuration, the simulation tool is mainly applicable for pre-
dictive analyses of DVE interactions and for studying hypothetical scenarios for
design and safety assessment purposes. However, once the simulation is validated
and streamlined by means of fast running parametrical expressions and specific al-
gorithms, it could be implemented in vehicle technology for real-time assessment
of DVE interaction. It could therefore be applied for detecting and anticipating po-
tential risky behaviours and dangerous conditions, and preventing them by either
suggesting possible driver intervention and/or by taking over control of the vehicle,
recovering the normal situation or containing the consequences of a incident.

References
Cacciabue, P.C. (2004). Guide to Applying Human Factors Methods. Springer-Verlag,
London.
Hollnagel, E. (1993). Human Reliability Analysis: Context and Control. Academic Press,
London.
Hollnagel, E. and Woods, D.D. (1983). Cognitive Systems Engineering: New wine in new
bottles. International Journal ofMan-Machine Studies, 18, 583-606.
 19. Simulation of Driver Performance 375

Kirwan, B. and Ainsworth, L.K. (1992). Guide to TaskAnalysis. Taylor and Francis, London.
Kondo, M. (1953). Directional stability (when steering is added). Journal of the Society of
Automotive Engineers ofJapan, 7(5/6),104-106, 109, 123, 136-140. (In Japanese.)
McRuer, D.T., Allen, R.W., Weir, D.H. and Klein, R.H. (1977). New results in driver steering
control. Human Factors, 19,381-397.
McRuer, D.T., Graham, D., Kredel, E. and Reisener, W. (1965). Human pilot dynamics in
compensatory systems - theory, models and experiments with controlled elements and
forcing function variations. Report AFFDL-TR65-15. Wright Patterson AFB, OH.
McRuer, D.T. and Krendel, E.S. (1974). Mathematical models of human pilot behavior.
AGARDograph No. 188. NATO Advisory Group for Aerospace Research and Develop-
ment.
Michon, J.A. (1985). A critical review of driver behaviour models: What do we know? What
should we do? In L.A. Evans and R.C. Schwing (Eds.). Human Behaviour and Traffic
Safety (pp. 487-525). Plenum Press, New York.
Minsky, M. (1975). A framework for representing knowledge. In P. Winston (Ed.). The
Psychology of Computer Vision. McGraw-Hill, New York.
Moray, N. (1997). Human factors in process control. In G. Salvendi (Ed.). Handbook of
Human Factors and Ergonomics (pp. 1945-1971). Wiley, New York.
Neisser, U. (1967). Cognitive Psychology. Appleton-Century-Crofts, New York.
Newell, A. and Simon, H.A. (1972). Human Problem Solving. Prentice-Hall, Englewood
Cliffs, NY.
Rasmussen, J. (1983). Skills, rules and knowledge: Signals, signs and symbols; and other
distinctions in human performance model. IEEE Transactions on Systems, Man, and
Cybernetics, 13(3), 257-267.
Re, C. and Carusi, E. (2005). Driving stability correlated with automated in-vehicle systems.
In 24th Annual European Conference on Human Decision Making, Athens.
Rouse, W.B. (1980). Systems Engineering Models of Human-Machine Interaction. North
Holland, Oxford.
Salvucci, D.D. (2001) Predicting the effects of in-car interface use on driver performance:
An integrated model approach. International Journal of Human-Computer Studies, 55,
85-107.
Stassen, H.G., Johannsen, G. and Moray, N. (1990). Internal representation, internal model,
human performance model and mental workload. Automatica, 26(4), 811-820.
Sheridan, T.B. (1992). Telerobotics, Automation and Human Supervisory Control. The MIT
Press, Cambridge, MA.
von Neumann, 1. (1958). The Computer and the Brain. Yale University Press, New Haven,
CT.
Wiener, N. (1948). Cybernetics. MIT Press, Cambridge, MA.
Wickens, C.D. (1984). Engineering Psychology and Human Performance. Merrill,
Columbus, OH.
Wickens, C.D. (2002). Multiple resources and performance prediction. Theoretical Issues
in Ergonomics Science, 3,159-177.
Weir, D.H. and Chao, K.C. (2007). Review of control theory models for directional and
speed control. In P.C. Cacciabue (Ed.). Modelling Driver Behaviour in Automotive En-
vironments. Springer, London.
 VIII
Simulation of Traffic
and Real Situations
20
Real-Time Traffic and Environment
Risk Estimation for the Optimisation of
Human-Machine Interaction
ANGELOS AMDITIS, ARIS POLYCHRONOPOULOS
AND E. BEKIARIS

20.1 Introduction
The development of next generation driver-vehicle interaction systems should be
focused towards obtaining a safe and sustainable mobility, with the aim to half
the number of road accidents as proposed by the European Commission (2001).
Mobility should be promoted in the future towards 'intermodality' in order to
reduce traffic congestion and to optimise travel planning; however, towards this
aim there is an increasing demand for on-board information systems. These needs
together with the demand for new on-vehicle support and services and the need of
the users to be connected to their own information cell (mobile phone, PDA, etc.)
will unavoidably increase the number of interaction of the driver with the vehicle
thus raising the potential risk of driver's distraction and fatigue, which are among
the main causes of road accidents.
While conventional vehicle safety measures (seatbelts and airbags) have con-
tributed significantly to the reduction of fatalities in the last decades, their safety
contribution is reaching its limits and currently further improvement is difficult to
be achieved at a reasonable cost. Today, the development of new advanced driver
assistance systems (ADAS, e.g. collision avoidance, lane-keeping aid and vision
enhancement systems) offers great potential for further improving road safety, in
particular by means of mitigating driver errors. In addition, the number of in-
vehicle information systems (IVIS) increases rapidly in today's vehicles. These
systems have the potential to greatly enhance mobility and comfort (e.g. navigation
aids and traffic information systems, media players, web browsers, etc.), but at the
same time increase the risk for excessive and dangerous levels of inattention to
the vehicle control task. Furthermore, many IVIS functions are today featured on
portable computing systems, such as PDAs or advanced mobile phones, which are
generally not designed for use while driving. In the near future, many of these func-
tions could be expected to be easily downloadable from a remote service centre,
directly to the vehicle or the nomadic device.
This variety of systems and functions that interact with the driver in one way or
another leads to a number of challenges, both technical and human factors related,
for the designer of the future automotive HMIs. These challenges include the HMI

379
380 Amditis, Polychronopoulos and Bekiaris

design for the individual systems as well as the question as to how to integrate a
range of different systems into a functioning whole with respect to their interaction
with the driver. Another challenge concerns how to best exploit the technological
possibilities of adaptive interfaces in the automotive domain.
The latter is the main proposition that the paper deals with. It primarily addresses
the adaptivity features of human-machine interaction related to the external traffic
and environmental conditions. In particular, the environment is modelled and risks
are calculated based on the current macroscopic driver's behaviour. A real-time
supervision system is being developed and presented, which detects, analyses and
assesses the environment. The outcome is a calculated level of risk on several
driving scenarios or states of the environment. According to the risk level, a mes-
sage can be presented to the driver or not, enabling the aforementioned adaptivity
features of integrated HMI systems.
In the sequel, two different use cases are presented that represent two different
systems and concepts: the AWAKE system, in which the traffic and environment
supervision is used to adapt a safety application to the environment - an adaptive
hypovigilance driver warning system (DWS) - and the AIDE system that offers
an adaptive integrated HMI solution for conflicts between on-board messages that
compete in order to be presented to the driver. AWAKE was the first attempt
to calculate an overall level of traffic risk, whereas AIDE proposes a complete
algorithm that covers most of the critical traffic scenarios.

20.2 The AWAKE Use Case - Adaptation of a Driver

Hypovigilance Warning System
20.2.1 AWAKE System Overview
Recent research indicates that driver hypovigilance is a major cause of road acci-
dents (Horne and Rayner, 1999). Accidents related to hypovigilance are generally
more serious than accidents related to other causes (alcohol, speeding, right of
way refusal) because a drowsy driver is not likely to take an evasive action prior
to a collision. If a cruise control is used, the vehicle may keep its speed until a
major impact. Reduction of collisions due to driver hypovigilance may contribute
significantly to a reduction of traffic crash losses. Drowsiness-related accidents
usually occur during late-night hours, with a smaller peak in the mid-afternoon.
Young drivers have no increased risk during the afternoon, while drivers over 45
have fewer night time crashes, and are more likely to have such crashes during the
mid-afternoon. It is also estimated that drivers younger than 30 years (constituting
one fourth of licensed drivers) account for almost two thirds of drowsy driving
crashes. These research results led to the need for the development of a system for
monitoring the driver's state: the AWAKE system (www.awake-eu.org).
Additionally, apart from normal drowsiness, the problem might be also directly
related to the introduction of various ADAS in the next years in the market. Auto-
mated or even assistive driving systems may also induce fatigue and stress to the
 20. Real-Time Traffic and Environment Risk Estimation 381

driver caused by prolonged driving under monotonous driving conditions. Stress

due to overload of information may also lead to fatigue. Therefore, the occurrence
of driver hypovigilance might well increase due to the introduction of ADAS.
However, at the same time ADAS technology may as well provide solutions to this
problem.
The objective of the AWAKE project was the development of an unobtrusive and
reliable in-vehicle system to monitor the driver and the environment, for real-time
detection of hypovigilance, based on multiple parameters. Continuous, instead of
discrete, event-related driver monitoring, effective system personalisation based
on driver characteristics, and consideration of the actual traffic situation enhances
the reliability of the system and minimises the false alarm rate. In case of hypovig-
ilance, the system provides an adequate warning to the driver. Several warning
levels were used, depending on the estimated level of driver's hypovigilance and
the estimated level of traffic risk. The system was designed for highway driving
and should operate reliably and effectively in all highway scenarios. The system
consists of the following modular components (Amditis et aI., 2002b):

Hypovigilance diagnosis module (HDM) that detects and diagnoses driver hy-
povigilance in real time. Based on an artificial intelligence algorithm this module
fuses data from on-board driver monitoring sensors (eyelid and steering grip data)
and data regarding the driver's behaviour (lane tracking, gas/brake and steering
position data).
A traffic risk estimation module (TRE) that assesses the traffic situation and
the involved risks. It matches, following a deterministic approach, data from an
enhanced digital navigational map, a positioning system, anti-collision radar, the
odometer and a driver's gaze direction sensor. This module is not designed as a
complete new system to estimate traffic risk, but rather as an expert combination
of existing ADAS technology. The output of this module was used by the HDM
to re-assess the state of the driver and by the DWS to determine the adequate
level of warning.
A DWS that uses acoustic, visual and/or haptic means. The module uses inputs
from the HDM and the TRE to determine the adequate warning level for a certain
situation.

20.2.2 Traffic Risk Estimation in AWAKE System

The purpose of a TRE module is to assess the traffic situation around the vehicle and
thus, the involved risks for the driver and passengers. The information computed
by TRE is communicated with a number named RLI (risk-level index), in which
there is information about the severity of the situation, deriving from the situation
itself. In on-board system architecture, the TRE module can be regarded as part of
the on-board information system. The TRE module generates useful information
to adequately prioritise the information flow towards the driver in view of the
surrounding traffic situation. Moreover, the computation of a general risk level can
also be used for other applications, e.g. for tuning of the warning strategies to the
382 Amditis, Polychronopoulos and Bekiaris

specific situation. In the AWAKE project, the evaluation of a general risk index is
used as input to the warning system. In summary, risk is a function of
the state of the vehicle (its dynamic condition described by speed, acceleration,
etc.);
road scenario (type of road: motorways, urban roads, etc.);
traffic scenario (complexity, speed, density);
environment (weather conditions, visibility);
driver conditions (perception of the surrounding environment, distraction).
A complete and integrated TRE system would be very complex and was far
from an industrial exploitation phase in AWAKE. It was not an objective of the
AWAKE project to develop new sensors to create such advanced system. However,
the concept of the TRE system is modular and can therefore be easily adapted to
the sensor system of the vehicle in which it is installed. TRE module consists
of four main blocks (Damiani et al., 2003). The first unit, sensors, includes both
the sensors that are standard available in the car, and which are shared with other
on-board functions (ABS, etc.), and specific sensors added to the vehicle for the
TRE module. The second unit, scenario assessment (SA), integrates all information
coming from the sensors. The third unit, warning strategies, identifies discrete risks
and their risk level, while the fourth unit, risk-level assessment (RLA), integrates
the discrete risks into a unique overall RLI.

20.2.3 The Scenario-Assessment Unit

The goal of an SA unit is to collect and process information from all sensors
and to make this information available to other units. In particular, it collects,
synchronises, filters and integrates the raw data from the sensors. Moreover, it
provides additional information derived from integration of sensor information
and combination with previous information (e.g., the gaze sensor provides angles,
from which may be calculated what the driver is actually looking at; the vehicle
trajectory is calculated from current position - calculated from sensor inputs -
in combination with previous positions) and describes the situation (scenario).
Different sensors have different sample frequencies and may not be synchronised.
The module has to manage these time and frequency differences in the flow of
incoming data, to match and translate the data into a flow of descriptions (pictures)
of the situation around the vehicle, and of the (dynamic) situation of the vehicle
itself, with a typical update rate of 10Hz. The unit has an open and modular
structure, which allows both different combinations of sensors, according to the
sensor availability in the vehicle in which the TRE module is installed, as well as
easier future addition of newly installed sensors.

20.2.4 The Warning Strategies Unit

The warning strategies unit receives a high-level situation description in terms of
environment identification and physical data. It produces a set of warning levels
 20. Real-Time Traffic and Environment Risk Estimation 383

TABLE 20.1. Definition of warning level index.

Warning level Situation

WLI No danger (normal situation)

WL2 Attention required (cautionary)
WL3 Danger (imminent risk)

that represent the risk level for each single event. The warning strategies unit
employs several algorithms that implement the specific warning criteria. As output,
a warning level index is generated as described in Table 20.1.

20.2.5 The Risk-Level Assessment Unit

The RLA unit receives a set of warning levels (one for each warning event that
is taken into consideration) and computes the final RLI. This value describes
the risk level of the overall situation and is also an output of the TRE (as are
the warning levels of the warning strategies unit). The RLI implicitly contains
information about the time factor of a dangerous situation and about the severity
of the potential damage. As mentioned before, the RLA unit receives a set of
warning levels (one for each warning event that is taken into consideration) and
computes the composite RLI. This value describes the risk level of the overall
situation and it is an output of the TRE module as well.
At the actual stage of work progresses, a scale for the TRE module output is
defined and proposed in which four levels of risk are distinguished, as described
in the previous table. Based on a survey of the state of the art of these topics (e.g.
Montanari et al., 2002), the messages to be given to the driver were divided into four
main groups depending on their priority level. This approach to assign four levels
of priority to the messages has been the starting point for defining the four levels
of risk as output from the RLA unit. According to Table 20.2, there is a normal
situation when no risks are identified and the corresponding risk level is given the
value of 1. The cautionary case is divided in two distinct levels, slight cautionary
and severe cautionary. Because of this distinction four (instead of three) risk-levels
emerge. Risk level 4, the imminent case, represents the most dangerous situation
and corresponds to a red alarm. This requires an immediate driver reaction. A
special problem is how to assess the overall traffic risk, i.e. how to reduce different

TABLE 20.2. Level of risk.

Risk level Associated
index Description colour

1 Normal situation (no risk estimated, scenario identified, factor detected) Green
2 Slight cautionary case (single scenario or factor) Yellow
3 Severe cautionary case (combination of scenarios and factors) Orange
4 Imminent case (highest risk) Red
384 Amditis, Polychronopoulos and Bekiaris

traffic risks to the same denominator. For this purpose, a method was proposed by
Damiani et al. (2003) for the classification of critical traffic scenarios according
to the associated level of risk. The more promising approach seems to be a mix
of a simple rule-based system with a weighted sum approach, which assigns a
pre-defined value to each single warning originating from a specific function. In
particular, the proposed rules are as follows:
Normal situation - If the levels of warning coming out from the different functions
are all at their own minimum value, then the output will be always 1.
Intermediate cases - In order to distinguish between levels 2 and 3, the criteria are
based on a sum of weighed values (a particular pre-defined value is assigned to
a warning level originating from a specific function).
Imminent case - As soon as a maximum level of warning is provided by one of
the functions, the output will be always 4.

20.3 The AIDE Use Case - Optimisation of the In-Vehicle

Human-Machine Interaction

The general objective of the AIDE (adaptive integrated driver-vehicle interface)

project (www.aide-eu.org and Engstrom et al., 2004) is to address these challenges
by means of tightly integrated interdisciplinary work, involving leading human
factors as well as technical expertise. Following are the specific topics addressed
by the project:

The development of a detailed understanding of the behavioural effects of inter-

acting with different types of in-vehicle functions.
The design of human-machine interaction (HMI) strategies for ADAS that max-
imises the safety potential of the systems.
The design of HMI strategies for IVIS that minimise the added workload and
distraction to the driving task.
The development of strategies for integrating multiple in-vehicle functions into
a coherent interface towards the driver.
The development of technological solutions, e.g. multimodal input-output de-
vices, driver-vehicle-environment monitoring techniques and an underlying
software architecture that supports this.
The development of solutions for the safe integration of nomadic devices into
the driving environment.
The development of valid and cost-efficient methods for in-vehicle HMI evalu-
ation with respect to safety and usability.

In the above context, driver-vehicle-environment (DVE) is modelled in order

(a) to understand driver behaviour in presence of ADAS and IVIS and (b) to esti-
mate in real time the current state of the DVE components. While (a) is not under
the scope of the paper, (b) is the main focus of the presented work. DVE real-time
modelling includes the development of a perception system that extends existing or
 20. Real-Time Traffic and Environment Risk Estimation 385

under development frameworks for safety of comfort applications, e.g. Tatschke

et al. (2006) in order to include driver's information to the perception architecture.
The DVE module estimates the state of the DVE individual components and cal-
culates level of risks as higher levels of abstraction for direct exploitation from the
HMI component. The following components are included:

(a) Traffic and environment risk estimation (vehicle, traffic, environment and their
relationships) - TERA
(b) Cockpit activity assessment (driver's activities not related to the driving task) -
CAA
(c) Driver availability estimation (driver's activities related to the driving task) -
DAE
(d) Driver state degradation (driver's physical ability) - DSD
(e) Driver characteristics (driver's individual behaviour and preferences) - DC

The (a) component is a successor of TRE module, namely the TERA module and
it is described in details in this paper.

20.3.1 Overview
Traffic and environment risk assessment (TERA) in AIDE monitors and measures
activities outside the vehicle in order to assess the external contributors to the en-
vironmental and traffic context and also to predict the driver's intention for lateral
manoeuvre. Existing sensors used by collision-warning (long-range radars), lane
departure warning (cameras), blind spot warning (cameras), maps and position-
ing systems combined with a table of corresponding roadway characteristics and
vehicle inertial sensors could be used to help understand the environment outside
the vehicle and adapt the HMI accordingly. The supervision algorithms collect all
available raw (e.g. radar signals) and processed data (e.g. tracked object list). The
role of TERA is threefold:

to calculate in real time a total level of risk related to traffic and environmental
parameters;
to calculate environmental and traffic parameters according to the requirements
of the other DVE modules and/or warnings if a function is absent in the vehicle
(e.g. the collision warning function could be produced from TERA if a radar is
available );
to estimate the drivers' intention (e.g. for manoeuvre of a possible lane change).

The above analysis is applicable to all possible parameters:

time to collision in longitudinal control systems;

predicted minimum distance and time to go in dynamic systems;
speed when approaching a curve;
environmental profile;
road profile.
386 Amditis, Polychronopoulos and Bekiaris

Radar
object hSl, wammg
Flag, rrc, TH TEM algorhh rns ,
,~--------.,I
,
..
: TE RA Output :

Inertial TERA ............_---

, othertoDVE ,,
I
I
I

V,IOCdy, Sln nn;,

: modules :
Pedal, . bbnhr. I--- (lntemal I ,
gear, yawRate parameters) I J,
Wipe rs , lights
Drive, Inte ntio n for la l.,aI
1----------"1,
.. man oe uvre detection
,,
,,
!
Frontal Cam'n11
Offn l, he.dl ng,
Warning.ag , CO.
t- Driver intentio n
(to be joined Wll:h TER A output ,,
Lane mart..lngs
RIsk
eM detecti on) to
leA and I
,
Assessme nt I
Lo.o p....'o. }- (Ded.on)
app l,catJons I
Risk
came'. I
functions I
ObJeC1 hst. Waml(lg I
FlaQ.lane ChanQ8 _____ _ ___ __ JI
- '- '- '- '-'- .I
I

I
Blind $pO'
carner. ~
I "_. _. _. _. _... !
I--
Cartographic da la ban and
positioning system: Speed
hmlt , road type .
--
Curvalure. inlerse ction
l.t:!P Lann . Pnonty, Pol, Sl~' .......

FIGURE 20.1. TERA module functional architecture.

20.3.2 Architecture
The role of TERA is represented by the two physical blocks, namely the risk
assessment and the intention detection. In Fig. 20.1, the connections with the
other components are indicated in the vehicle environment. The physical inter-
faces are out of the scope of this document (CAN , TCP, etc.). An adequate soft-
ware module is under development, to combine all different pieces of informa-
tion from the various subsys tems in a coherent whole, running on an on-board
computer.
Input sensor array includes input control sensors (e.g. steering wheel angle sen-
sor, pedal position sensor), enviro nment sensors (radar, laser, IR, etc., but also GPS
and digital maps) and vehicle dynamic state sensors (speed sensors, accelerome-
ters, yaw rate sensors).

20.3.2.1 Relevance to the AIDE Use Cases

Starting from the AIDE design scenarios, a scheme for adapting a vehicle HMI
to a high traffic or environment risk situation is presented. Two different cases
(in terms of the cause and not of the proposed solution) of HMI adaptation are
considered accordingly to TERA module's outputs:

High traffic risk situations - High traffic risk value should signify a quite dan-
gerous situation where a warning might have already been generated or the
possibility to be generated is high. In those cases , adaptation can only have the
 20. Real-Time Traffic and Environment Risk Estimation 387

meaning of a stricter prioritisation by the leA in order to let only warnings to be

presented (and even only one warning in order to minimise possible distraction
effects from other messages in a difficult driving situation). Thus, high traffic
risk value affects only output messages' scheduling (delay or cancel).
High environment risk situations - High environment risk value signifies an
increased workload situation where the reaction time of the driver might be re-
duced due to road characteristics or bad weather conditions and thus we propose
to present important information or warnings earlier or with greater intensity. A
modality adaptation could also be the case if a recommended alternative modal-
ity is available for an output message. In a heavy rain situation, where the audio
messages are not so well received due to the surrounding noise, a modality al-
teration to visual could occur. Thus, high environment risk value can affect the
presentation of both warnings and important output messages.

In addition, to the two former subcategories ofTERA-detected conditions, there

is also a third output of this module which will affect the AIDE HMI. This is the
detection of the driver intention to perform a lateral manoeuvre.

Lateral manoeuvring intent - In the envisioned HMI adaptation functionality,

is generally of interest that the intention detection will be able to discriminate
between a simple manoeuvre and a high-demanding manoeuvre in order to
treat accordingly low- and high-priority information scheduling. Finally, it can
be noted that in the general envisioned HMI functionality, detecting on-going
manoeuvres, is considered a case of driving demand detection. The intention
detection is presented in Section 20.3.4.

20.3.2.2 Description of Environment

The challenge in recent years is the environment recognition all around the subject
vehicle so as to be able to prevent risks (collision, lane/road departure, etc.) in
the longitudinal, rear and lateral field. Traffic and environment risk within AIDE
is perceived as a major contributor to the driver's workload. The elements that
describe the scenario that is assessed through the TERA module are as follows:

(a) the road infrastructure consisting of the lanes, the road borders and the infras-
tructure elements (e.g. speed limit, traffic signs);
(b) the subject vehicle and its dynamics;
(c) the moving and stationary obstacles;
(d) the traffic flow representing the number of obstacles ahead or/and their dis-
tances from the ego-vehicle;
(e) environmental parameters.

Using the vehicle data (e.g. yaw-rate sensor) and the data coming from 'exter-
nal sensors' (radar and possibly camera), it is possible to compute and assess the
subject vehicle's trajectory. The method that is used within AIDE TERA module
is the scenario representation by analysing the following state elements (Poly-
chronopoulos et al., 2005):
388 Amditis, Polychronopoulos and Bekiaris

(a) Object State = the dynamics of the objects in the longitudinal field.
(b) Subject State = the dynamics of the subject-vehicle.
(c) Road State = the geometry of the lanes and road borders.

The vehicle's present curvature (Co) is represented by the following formula:

Co = Q/v, where Q is the yaw-rate value, v is the vehicle speed and Co is the
inverse of the radius of the curvature covered by the vehicle (Co = 1/R). The
more sophisticated is the model that is used, the more precise the vehicle motion
assessment will be. All elements in the proposed approach can be mathematically
represented by a state vector and are briefly described below:

(a) The road borders and lanes are described by a clothoid model (Kirchner and
%
Heinrich, 1998): Y (x) = co~ + Cl + Yo, where Yo is the offset from the
ego-vehicle's position, Co is the road curvature and YOl the rate of the curvature.
Different offsets YOl and YOr represent the left and the right border locations.
Thus, the following state vector can describe the road (or the lane): XRB =
(co Cl YOl YOr)T. The measurement space includes higher level parameters
from a camera (and its image processing unit for lane detection), map and
positioning data and inertial sensors (odometer, yaw rate sensor).
(b) The state of the subject vehicle (SV) contains kinematics, attributes and prop-
x
erties. A typical state vector of the former case is sv = ( V awe)T, where
e
V is the velocity, aas the tangential acceleration, is the heading and to the
yaw rate.
(c) The state of the obstacles contains also kinematics, attributes and properties
such as: Xo = ( x Y Vx Vy ax ay W h )T. Here, (x, y) are the Cartesian co-
ordinates in a local coordinate system (i.e. the subject's vehicle coordinate
system) and V, and a is the relative velocity and acceleration respectively in
the two axis. Wand h refer to the properties of the tracked obstacles.

The measurement space is produced by the radar signals corresponding to mov-

ing obstacles due to the Doppler effect. Other signals coming from the vehicle bus
(CAN bus - controller area network) or from the map database assist to the correct
interpretation of the scenario. These signals are raw data communicated through
the vehicle network:

(a) Road data - Type of road context (country, urban, peri-urban, highway, urban
highway), priority level of the road, number of lane, speed limit, presence of
school, etc.
(a) Vehicle data - Blinkers status, wipers' position and light position.

20.3.3 Algorithm for Risk Assessment

20.3.3.1 Rule-Based System Employed for TERA Algorithms
TERA module expected condition detection involves detection of the following
conditions:
 20. Real-Time Traffic and Environment Risk Estimation 389

Driver is approaching a dangerous curve with high speed - A prediction of the

risk involved in terms of the vehicle leaving its lane, or road, in the curve due to
over speeding with respect to the road's curvature.
Risk of frontal/lateral collision - Frontal and lateral obstacles presence, and
distance and acceleration estimation should lead to a decision on the potential
unsafe driving situation with respect to the ego-vehicle's path against surround-
ing vehicles' estimated path.
Risk of road/lane exit - Risk involved with respect to the lateral position of the
vehicle in comparison with the road/lane limits.

On the basis of these three basic conditions, TERA risk assessment produces a
set of traffic risk functions that take into consideration not only this detection of
discrete dangerous situations around the subject vehicle but also the traffic and the
environment context where these situations occur. However, in order to ensure the
correct system's output in case of a dangerous situation, this conditions' correlation
with the traffic and environment context takes place only when the basic conditions
are not classified as imminent.
In the following, the rule-based approach adopted for TERA algorithms is illus-
trated explaining the functions' derivation. In each of the TERA functions, the first
coefficient reflects the main condition to be detected, where the second coefficient
reflects the risk contributor derived from specific characteristics (parameters) of
the current driving scenario. Each coefficient inside the second coefficient repre-
sents a sub-condition (subsidiary condition) and is multiplied with a number W
between 0 and 1 (confidence weight) where Lm Wi == 1 and m is the amount of
selected sub-conditions (if main condition NOT imminent):
TERA F unct ion == (main-condition) x {[(sub_condition 1) x
wl]+[(sub_condition2) x W2]+ ... [(sub-condition) x wm ] }

where (float) TERAFunction E [0,1] and each of the main conditions and their cor-
responding sub-conditions can either have deterministic (wipers status) or fuzzy
(TTC) values. In the first case, variables that are involved in the conditions or
sub-conditions' derivation are modelled by membership functions (trapezoidal
or Gaussian) and thus, the operations among variables are replaced from op-
erations among fuzzy sets. In the second case, relevant variables have scalar
or Boolean values. In addition, a sub-condition can also be the logical output
of two or more conditions following the binary aggregation according to the
Boole algebra. Weights give the possibility to externally control system's out-
puts. Moreover each condition can influence in a different way each function
which gives the system a more realistic-intelligent overview of the current see-
nano,
Thus, the enhanced rule-based approach adopted for each of the TERA traffic
risk functions is based on estimating one main condition (high level situation
description), which is then modulated by a factor representing the confidence
assigned to the main condition detection. This factor consists of a weighted sum
of the sub-conditions considered to influence the specific main condition. Such
390 Amditis, Polychronopoulos and Bekiaris

sub-conditions may involve consideration of the data coming from the vehicle
inertial sensors (including environmental information) and the data coming from
the map and positioning system combined with a table of corresponding roadway
characteristics in order to create a more global picture of the environment outside
the vehicle within the current driving scenario. However, it should be noted once
again that this approach is valid only if the main condition has not been classified
as imminent.
The main conditions to be detected lead to the three following traffic risk
functions:

Risk of approaching a curve with high speed (DangCurveAppr).

Risk of frontal/lateral collision (RiskLFColl).
Risk of road/lane exit (RiskRLExit).

First weights will be defined from experts' opinions and previous experience
in the field (Bekiaris and Portouli, 1999; Damiani et al., 2003). TERA function
concerning environment risk assessment will be similar to the way that traffic risk
is calculated.
In the second step, an overall risk assessment function associates these higher
level functions with each other and eventually produces the two risk functions
required by the ICA in order to adapt the system's HMI with respect to traffic and
environment risk estimation. This association of the individual higher level risk
functions adopts a weighted-aggregation approach based on experts' opinions and
TERA team's previous experience in the field.
The final overall traffic risk function (TrRisk) can be thus derived using a
weighted sum of the three internal risk functions:
TrRisk = (DangCurveAppr > thres1) x WI + (RiskLFColl > thres2) x W2 +
(RiskRLExit> thres3) x w3
First weights will be defined from experts' opinions and previous experience in
the field and modulated in accordance with real data sets tests.

20.3.3.2 Main Traffic Risk Condition Detection

The risk assessment supervision algorithms collect all available raw (e.g. radar
signals) and processed data (e.g. tracked object list) from the sensors in terms of
the traffic and the environment parameters. The risk assessment produces a set of
basic risk functions representing an interpretation of the current driving scenario
in terms of discrete dangers around the vehicle. The types of possible situations
are frontal (lateral/blind spot) collision, lane/road departure and dangerous curve
approaching situation. The functions used to evaluate each of the above situations
are based on deterministic consideration of the road scenario. The risk value is
extracted by comparing the value of relevant parameters with pre-defined fixed or
adaptive thresholds. For each function, three modalities are defined: information
and imminent and an intermediate cautionary case.
 20. Real-Time Traffic and Environment Risk Estimation 391

20.3.3.2.1 Risk of Frontal/(Lateral) Collision

The frontal/lateral collision risk function has the aim to evaluate the risk due to
an obstacle present on the host-vehicle trajectory. Such a function uses a long-
range radar to assess the distance, the velocity and the angular position of the
objects ahead or aside. The fundamental point is to understand if one (or more)
object(s) detected by the radar in front of the vehicle is dangerous, and in case
it becomes an obstacle, or not. In order to assess this event, the object has to be
inside the driving path (a path wide 1.5 to 2 m, more or less the same width of
an ordinary sedan): If this is the case, it is considered as a potentially dangerous
obstacle. In the method used by conventional ACC systems, one main assumption
is made about the motion of the objects (deterministic method which ignores the
time parameter): The host-vehicle obstacle is regarded as moving with constant
speed (straight uniform motion).

20.3.3.2.2 Criteria ofAssigning the Level ofRisk

The common measurements concerning safety distance are calculated according

to the criteria of 'time-to-collision (TTC)' and 'headway':

TTC is the time that results from the distance ~x (between leading and rear
bumper) from a leading car (carleading) to the own car (carsystem), divided by the
difference of velocity (vsystem - Vleading) between both of them

tTTC == ~x I.
I vsystem - Vleading
Headway is the time that results from the above-named distance ~x and the
velocity vsystem of the system car

~x
tHeadway == - - .
vsystem
The corresponding warning distances are calculated with a predetermined time
(tTTC and tHeadway) for several dangerous situations (cautionary and imminent). The
larger calculated warning distance is defined as criterion of the system activation.
The determination of the specific thresholds of TTC and headway for both danger-
ous situations is accomplished deriving from NHTSA (see below). The imminent
case applies to

TTC between 3 and 5 s or less;

headway between 1 and 1.5 s or less.

The cautionary case applies to

TTC is larger than in the imminent case but not larger than 10 to 14 s (5 s ~
TTC ~ 10 to 14 s);
headway is larger than in the imminent case but not larger than 2 to 3 s (1.5 s ~
headway ~ 2 to 3 s).
392 Amditis, Polychronopoulos and Bekiaris

A
2

r
tY\
v..
h Ay\ ~ iJAlI
o

-1
\A f ~ ~ ( .....
'y~

-2 ~ I
~
-3
~
-4

-5
o 200 400 600 800 1000 1200 1400 1600 1800 2000

FIGURE 20.2. Lateral velocity (m1s) evolution in time in a given scenario.

20.3.3.2.3 Risk ofLane/Road Departure

The lane/road risk function evaluates the risk due to unintentional lane departure
or in a more general way the risk due to a wrong lateral behaviour of the driver
in the road (i.e. travelling between two lanes). The frontal camera is the main
sensor for this function but information about speed and use of indicators are also
considered.
The strategy that seems to be the nearest to driver's needs (to have a proper and
constant reaction time) is 'Time to Line Cross': when the time to cross the line
decreases under a threshold time, a warning is issued. These computations take
into account the lateral speed of the vehicle. If the lateral speed is low, the car
is lined up to the road. Even though the car is close to the border, this condition
is not dangerous; when the lateral speed increases quickly, the situation becomes
critical. Thus, detected condition follows the conjunction:
if (lod > threshold) then activate warning, where lod == left offset derivative
In order to guarantee a suitable driver's decision and action time, the system
can be perceived as too strict, increasing its risk level before the line is crossed.
In order to guarantee a good comfort level of driving the time to line cross pa-
rameter is set very low so that the risk level becomes imminent very close to the
marker. Threshold definition is subject to the experimental procedure where real
data values (plots of input data) will be compared with what was really happen-
ing in the driving field (real time video data). In Fig. 20.2, the deviation of left
lateral velocity is presented indicatively taking data from an image-processing
unit.
 20. Real-Time Traffic and Environment Risk Estimation 393

20.3.3.2.4 Risk ofApproaching a Dangerous Curve Too Fast

The dangerous curve approaching risk function has the aim to evaluate the risk in
case the driver is approaching a curve too fast. The followed approach is similar
to the frontal/lateral risk function one, but in this case the obstacle ahead is the
curve, which is static and so only the motion of the subject vehicle must be taken
into account.
The goal of this method is to arrive at a certain time with a determined velocity
(the right one to route into the bend). This could be considered a reference speed,
depending, of course, on the type of vehicle and curve (and in particular on its
curvature). Using formulas, the reference speed (v f) is v f == J alat x R, where alat
J
== maximum lateral acceleration requested for the vehicle and R == o ' with Co ==
the curvature of the bend.
In this context, lateral acceleration values (for cautionary and imminent warning)
are stated by proper experimental phase, deriving the reference speed, which is
compared with the current one of the host vehicle. If VI 2: v., then the (longitudinal)
deceleration value to arrive at a certain distance dwarning from the curve with the right
velocity (namely the reference speed) is computed. If it is over a defined threshold,
the risk level is modulated accordingly. Thresholds' definition will be subjects of
experimental procedure. Other possible methodologies to compute the curvature
(however, not treated in this approach) are the B-Spline and the Bezier methods.

20.3.4 Algorithmfor Estimating the Intention of the Driver

The intention detection is carrying out a complementary task, which is the pre-
diction of the intention of the driver based on possible overtaking or lane change
manoeuvre. To detect a manoeuvre is rather simple by monitoring the steering
angle or the lateral behaviour of the vehicle. For example, if the derivative of the
lateral offset is estimated by a Kalman filter and it is compared with a threshold
then the manoeuvre estimator will detect all steering and corrective actions of the
driver. In Fig. 20.3, 600 successive system scans are plotted for such a detector
(0 - no manoeuvre, 1 left manoeuvre, -1 right manoeuvre); in this scenario only
one overtake between scans 690 and 830 takes place.
TERA module with its assessment properties uses a decision fusion system
(versus any other conventional rule based system) in order to detect lane changes
within a critical time before they occur. Thus, in order to solve the problem of
diagnosing a lane change, evidence theory of Dempster-Shafer (D-S) is applied
to realise the information fusion of multi-parameter in determining lane changes.
In the D-S theory (Shafer, 1976), the set of all possible outcomes in a random ex-
periments is called the frame of discernment (FOD), usually denoted by (). The (2)
subsets of () are called propositions, and probability masses are assigned to propo-
sitions, i.e. to subsets of (). The interpretation to be given to the probability mass
assigned to a subset of ()is that the mass is free to move to any element of the subset.
Under this interpretation, the probability mass assigned to () represents ignorance,
since this mass may move to any element of the entire FOD. When a source of
394 Amditis, Polychronopoulos and Bekiaris

I
1
1
I
1
1
I- manouver type I
I
I
1
1 1 I 1 1
0.8 - 1 I
r----------r---------~----------l--
1 1
I

1
1

1
--- --- ---- -----+----------
I
1 I 1 1 I
1 I 1 I 1

-----+----------
I 1 1 I
--- --- ----
1
0.6 - ~----------~---------~----------~--
1 1 1 1 I I
I 1 I 1 I I
I 1 I I I 1
L
I __________ 1
L _________ J I __________ 1 __ 1 _ ____ L1 __________
0.4 - --- --- I ----
~

1 I 1 I 1
1 1 1 I I I
1 1 1 1 I I
1 1 1 I I I

0.2 -
1 1 1 1
r----------~---------4----------~--
I 1 I 1
--- --- 1 ----
I
-----T----------
1

1
I 1 1 I 1 I
I 1 1 I 1 I

o
I I 1 I 1 1

1 1 1
- 1
--- "--~---- 1
----,-
1 1 1 1 I 1
1 1 1 1 I 1
1 1 I I I 1

-0.2 ---r-------
I

I
1
--~---------4---
I
I

1
-
I 1
----~----------~----------T-
I 1
1

I
------ -
1 1 1 1 I 1
1 I 1 1 I 1
___ 1 _______
L 1 _________ J 1___
__ L _ ___ 1__________ I __________ L1 _
- ------ -
~ ~

-0.4 1 1 1 1 I 1
I 1 I I I 1
I 1 I I 1 I
I I 1 1 1 1

-0.6 ---r-------
I I 1 1 1 I
- -r - - - - - - - - - -1- - - - ----~----------~----------T- ------ -
1 1 1 1 1 1
1 1 I 1 I 1
1 1 I I 1 I
___ 1 _______
L __ 1L _________ J I ___ _ ___ 1 __________ I __________ L_
1
-0.8 - ------ -
~ ~

I I 1 I 1 I
I I 1 1 1 1
1 I 1 1 1 1
1 1 1 1 1 1

-1 ---~------- --~-- ---- .i..; - ----~--- ------~- ---- --~-

300 400 500 600 700 800 900

FIGURE 20.3. Manoeuvre detection.

evidence assigns probability masses to the propositions represented by subsets of

q, the resulting function is called a basic probability assignment (BPA). Formally,
a BPA is function m: 2() --* [0, 1], where m (0) == 0 and LAce meA) == 1. Subsets
of () that are assigned non-zero probability mass are said to be focal elements of
m. The core of m is the union of its focal elements. A belief function, Bel (A), over
() is defined by Bel(A) == LBCA m(B).
In the case of TERA, the FOD is () == {lane change, no lane change}. It is critical
to determine a basic probability assignment when we use the D-S model to match
the information. Generally, the basic probability assignment is closely relative
to the data type and special objective. When we established the basic probability
assignment, we acquired knowledge from experts who got more useful information
ADAS systems. The source of evidence selected is

1. time to lane crossing (TLC);

2. lateral offset (lateral position of the vehicle with respect to the middle of the
lane);
3. derivative of the lateral offset (lateral velocity);
4. the difference between the curvature of the road and the curvature that the
vehicle is following, i.e. Co - (J) / V, where (J) is the yaw rate, Co is the road
curvature and V the velocity of the vehicle;
5. the product of the curvature of the road and the curvature that the vehicle is
following Co . V;
if the product is negative, then this is an evidence of a lane
change;
6. The gaze of the driver, if the driver is looking at the side mirrors of the car.
 20. Real-Time Traffic and Environment Risk Estimation 395

1 1 1
I-- ~I ~ ,.. - ~ - - - - - - - -1- - '" - ..,...._ _.__ - - _ _...+-.._._ -.............-+
I 1 1
I I 1
I 1 1

0.9 _______ 1 L
1 _ I

--------------- -------------------------
~

1 1 I
1 1 1 1 1
1 1 I I 1
I 1 1 I I 1
1 1 I 1 1 I
0.8 _ _ _ _ _ _ _ L
1
1

1
_ ______ L_____
1
J
I
L
I
L
I
_

1 I 1 1 1 1

,
I 1 1 1 I 1
I , 1 1 I 1

0.7 -- - - - - - - - - - - - - -,-- - - - - - - - -
I , 1
-----.-- 1
--1--------,--------.-------
I I 1

'E
Q) 1 I I 1 I 1
I 1 I 1 1 I
E I 1 1 1 1 1
c 0.6 1
- - - - - - - r - - - - - - - -,- - - - - - - - , - -
1 I
-r
1
----,--------r--------r-------
I I 1

0> 1 I I 1 I I 1
'00 1 I I 1 1 , 1
en 1 I I I , 1
Cl3
0.5 -------r------- ------,--------r--------r-------
1 I I I I 1
-j- - - - - - - - '1 - - - - -

~
1 , , I 1 I
1 , 1 I 1 1

:.0 1
1
1
1
1
,
I
I
,
I
I
1
Cl3
..c 0.4 1 1

a.
0 I 1
1 1

_ _ _ _ _ _ _ L1 1
, I
1 _ _____ L___
1
__J
1
_ LC
0.3 I 1 I 1 1
1
I
1
I
,
,
1
1
1
I
no LC
1 I I 1 1
union
-------:-----
I I I I

0.2 - - - - - - - ,

I
- - - - - - - -1- -

1
- - - - - -
:-----~---------~
1 1 I I
I I I

- - - - - - - \I - - - - - - - -,-I - - - - - - --------.------- -------\--------.-------

1 I 1

0.1 1 1 I I '
I I 1 I I
1 I 1 1 I
1 I 1 1 I
0
-0.6 -0.4 -0.2 o 0.2 0.4 0.6 0.8
m/s

FIGURE 20.4. BPA for the derivative of the lateral offset.

The basic probability assignment, for each evidence, is calculated through proper
trapezoidal fuzzy membership functions. An example is given in the following
figure for the derivative of the lateral offset. The figure shows that if the value of
the derivative is 0.2 rn/s then 0.9 is the BPA assigned for the lane change and 0.1
for its negation (Fig. 20.4).
Assume that belief function Bel(i) are assigned to independent sources of ev-
idence in same frame of discernment and the relevant basic probability assign-
ment is m i, ma. etc. Then according to Dempster's rule of combination, the new
belief function Bel(A) and basic probability assignment meA) may be yielded
via

20.3.5 TERA Implementation

Fig. 20.5 illustrates the process of input data, categorised in four categories,
exploitation by the traffic and environment risk supervision algorithm, whilst
Fig. 20.6 shows a more detailed data mapping into the three TERA traffic risk
functions.
396 Amditis, Polychronopoulos and Bekiaris

CAN External sensors Lanes

LateraIOffeset
velocity ObjectNumber
LaneWindth
yaw rate filtered radial velocity
Lanecurvature
steering angle ObjDistance
LaneCurvatureRate
wiper position ObjAngle
Type of Lane marking
(headlamps) (ObjClass,ObjID)
(neighbourlanes)

Raw data + processed data + tracking ~

functions already ready: ObjectPath(), ObjectClosest(), PathPrediction(),
ObjectSynthesis()

JJ
Figuring out the
traffic algorithm which will environment
provide the envisioned
traffic and environment
risk assessment Environmental Risk:
OveraliTrafficRisk: f(weather conditions, type
f(DangCurveAppr, of road), to be reflected in two
RiskFLColI, separate conditions:
RiskRoLExit) - LowVisibility
- Low Audiblityin the Cockpit
(dueto differentHMI adaptation
requested)

FIGURE 20.5. Available signals versus TERA functional requirements.

DangCurveAppr Risk of Fr/Lat Collision Risk of Road/Lane Exit

= f (...) = f( ...) = f( ...)

Lateral posotion vs.

Ego-vehicle vs road TTC (long&lat), number of
lateral acceleration,
curvature, obstacles around in a specific distance,
environmental conditions,
accel/deccel, steering angle, environmental conditions,
indicators' status type of lane
environmental conditions, ego-vehiclevs. speed limit, distanceto
marking,numberof lanes +
nextCurveApp, ego-vehiclevs Intersection or traffic sign, type of road,
neighbourlanes to the right!
speed limit, blind spot object vs. intention,maneuver
left
type of road detectionof front tracked obstacle

FIGURE 20.6. Available signals mapped onto TERA traffic risk functions.
 20. Real-Time Traffic and Environment Risk Estimation 397

.~ I.I>[ fun., TRA Tool ~- ~

, e-d _ _

...- _... ., .,
r
C.aIiIDl :

~o.,..,:
055

005 ...
FIGURE 20.7. A screenshot from 'Fuzzy TERA development tool 1.0 ' laboratory tool.

The algorithms both for risk assessment and driver intention have been im-
plemented in MATLAB and C++. A tool (fuzzy TERA development tool 1.0)
has been developed that loads data sequences and video files in a playback
mode, synchronises data , runs off-line the TERA algorithm and displa ys re-
sults in a graphical user interface. The tool work s both with real (recorded from
test cars and truck s) and simulated data and it is developed for evaluation pur-
poses. The tool (Fig. 20.7) shows internal traffic risk conditions' values, video
and mark s whether the driver is considered to be in high traffic risk situation
(three detection levels). The TERA traffic risk conditions' threshold s are man-
ually tuned by varying several thresholds and weights and comparing the algo-
rithm with recorded video s of real driving. The algorithm for the detection of
possible manoeuvres is integrated in the same GUI. The C++ version is in-
tegrated in an automotive proce ssing unit and will run in AIDE demonstrator
vehicles.

20.4 Conclusion s
The paper presented a novel approach on the real-time assess ment of the traffic
and environmental situation with respect to the risk that the driver is exposed
to. The work focuses on the descripti on of the relevant risk in a higher level of
398 Amditis, Polychronopoulos and Bekiaris

abstraction so as to be used in advanced driver support systems or adaptive warning

and intervention systems.
More specifically, the so-called TRE module, and its successor TERA module,
attempts to assess the criticality of the current traffic and environmental conditions
by calculating a level of traffic and environmental risk, respectively. The module
is supervising the surroundings using sensing input devices (active and passive
sensors, digital maps, etc.), establishes a view of activities, manoeuvres, locations
and properties of road elements, and from it assesses what is happening or going to
happen and the severity which events will occur. These events include the risk of
collision, risk of road/lane exit, risk of dangerous approaching curve and environ-
mental risks. Additionally, TERA, starting from surround sensors, estimates the
driver's intention for lateral manoeuvres. Therefore, TERA allows the HMI con-
troller and the applications to be adapted to the driving demand and the criticality
of the situation both in terms of traffic and external environmental conditions.
TERA module is developed as a first prototype by using a limited set of sensing
devices, but covering major possible causes of critical driving demand. Critical
real-world scenarios should be fed into the algorithm and test the initial hypoth-
esis made in order to re-work and tune the algorithms. TERA is based on fuzzy
evidential reasoning for the calculation of risk and situation criticality and on D-S
theory of evidence for the intention detection.

References
Amditis, A., Polychronopoulos, A., Belotti, F. and Montanari R. (2002a). Strategy plan
definition for the management of the information flow through an HMI unit inside a car.
In e-Safety Conference Proceedings, Lyon.
Amditis, A., Polychronopoulos, A., Bekiaris, E. and Antonello, C. (2002b). System archi-
tecture of a driver's monitoring and hypovigilance warning system. In IEEE Intelligent
Vehicle Symposium Versailles, France.
Bekiaris, E. and Portouli, E. (1999). Driver warning strategies Internal Deliverable ID4.1.
IN-ARTE Project TR4014.
Damiani, S., Antonello, C., Tango, F. and Saroldi, A. (2003). Functional specification and
architecture for the traffic risk estimation module. In Proceedings SICICA 2003, Aveiro,
Portugal.
Engstrom, 1., Cacciabue, C., Janssen, W., Amditis, A., Andreone, L., Bengler, K., Eschler,
1. and Nathan, F. (2004). The AIDE integrated project: An overview. In Proceedings of
ITS Congress, Budapest.
European Commission. (2001). White Paper on European Transport Policy for 2010: Time
to decide.
Horne, J. and Reyner, L. (1999). Vehicle accidents related to sleep: A review. Occupational
and Environmental Medicine, 56, 289-294.
Kirchner, A. and Heinrich, T. (1998). Model based detection of road boundaries with a laser
scanner. In Proceedings ofIEEE International Conference on Intelligent Vehicles, Vol. 1,
Stuttgart, Germany, 1998, 93-98.
Montanari, R. et al. (2002). COMUNICAR: Integrated on-vehicle human machine interface
designed to avoid driver information overload. In ITS Conference, Chicago, ILO.
 20. Real-Time Traffic and Environment Risk Estimation 399

Polychronopoulos, A., Andreone, L. and Amditis A. (2005). Real time environmental

and traffic supervision for adaptive interfaces in intelligent vehicles. In IFAC congress.
Prague.
Shafer, G. (1976). A Mathematical Theory ofEvidence. Princeton University Press, Prince-
ton, NJ.
Tatschke et al. (2006). ProFusion2: Towards a modular, robust and reliable fusion architec-
ture for automotive environment perception. In V. Jurgen and G. Wolfgang (Eds.). Ad-
vanced Microsystems for Automotive Applications 2006 (pp. 451-470). Springer, New
York.
21
Present and Future of Simulation
Traffic Models
FABIO TANGO, ROBERTO MONTANARI AND STEFANO MARZANI

21.1 Introduction
In transportation research, simulation technologies have acquired a huge relevance
since they permit to reproduce, under controllable conditions, different scenarios
with a growing degree of complexity. In this sense, a general trend can be antici-
pated here since it represents one of the backbone of this paper. The simulators of
traffic scenario are enlarging the numbers of factors to be considered, from the sole
link between vehicles and driving environment to a joint scenario in which drivers'
intentions, autonomous behaviours of different vehicles and adaptive technologies
(providing ad-hoc reactions according to specific driving conditions) are all to-
gether considered and computed. Therefore, the more the complexity grows, the
more the network of factors influencing the reliability of these scenarios becomes
articulated.
Proposing different working cases (some of them still under research), this paper
tries to describe the main technological features of the traffic simulator model, the
corresponding design approaches and the uses that these simulators offer towards
the research community.
According to the results of the surveys done while writing this paper, a key
point in which a reliable simulation is not yet completely achieved (even if many
benchmark initiatives are acquiring promising results) is a study of drivers' be-
haviours and prediction of drivers' status, especially in terms of workload. This
is a crucial challenge since once the level of workload will be properly and in
real time monitored, the researchers will have the chance to actually improve safe
systems, reducing the number (still too high 1) of vehicles' related fatalities. Since
the rationale behind a workload monitoring system seems to be mainly heuristi-
cally detected (Balaban et al., 2004; Dixon et al., 2005; Recarte and Nunes, 2003;
Schvaneveldt et al., 1998; Zhang et al., 2004; Zhang et al., 2004), simulation
plays a decisive role in their assessment and in the evaluation of the most suitable

1 From many accident statistics carried out at European project level (PREVENT, AIDE,
etc.), each year there are around 1,400,000 accidents, out of which 40,000 are fatalities.

400
 21. Present and Future of Simulation Traffic Models 401

counteractions (e.g. a delayed delivery of low priority information (low priority

delivering in case the driving situation is critical).
This paper is organised as follows: In the first part, an overview of the main types
of traffic simulators, including their relevant computing components and techno-
logical solutions, is reported. Later, many use cases are described and discussed.
The use cases depict the transition from the standard model of traffic simulators
to the complex ones where driver, vehicle and environment models are joined and
embedded. The last part is dedicated to the integration between the traffic simulator
complex models, the paradigm of distributed intelligence and multi-agent systems
approach.

21.2 Traffic Simulator

21.2.1 General Overview: A Survey of

Road Traffic Simulations
Hundreds of traffic simulation models are available nowadays. Some of them are
quite effective and widely applied by the transportation research community for
research in different domains, from human factors to advanced engineering and
from road traffic to driver's training.
Recent studies on the competences and facilities of a sample of 34 human-
machine interaction R&D laboratories and companies throughout Europe found
that most of them (c.a. 45% of the sample) have adopted a driving simulator to
conduct specific tests on driver behaviour. For example, TNO (http://www.tno.nl)
is set up with a simulation platform reproducing a BMW or a DAF truck cabin
with a 120 visual scene and six DOF motion. The Competence Centre Virtual
0

Environment of Fraunhofer Institute (www.ve.iao.fhg.de) is equipped with three

simulators, from low to high fidelity driving environment's definition. Many other
research institutes and firms, such as Renault (France), Cidaut (Spain), ICCS 1-
SENSE (Greece), have embraced these kind of technology (Ferrarini, 2005; Minin,
2005).
Among the others, the topic of representing traffic congestion in virtual simula-
tion has a particular relevance since it is one of the causes of lost productivity and
decreasing standard of living in urban settings, including of course safety. How-
ever, limited by some constraints, especially the computational power availability,
current traffic simulators are not still completely realistic, which is the eternal goal
of simulation (Champion et aI., 1999).
The traffic simulation deals with dynamic problems in complex environment
that cannot be easily described analytically (Lieberman and Rathi, 1997). These
problems derive from the interaction of several components of the system (namely
its entities), which aim at reproducing (and in a certain way at 'mimicking') the
behaviour and the interactions between the real traffic actors (cars, trucks, etc.)
as accurately as possible. The simulation produces both statistical results and a
visual rendering. The statistical results are numerical, i.e. they give quantitative
402 Tango, Montanari and Marzani

and qualitative information referring to the evolution of the simulation. With this
knowledge background, a visual representation is created, and it provides an idea
regarding the state of the simulated environment. The main fields of applications
for a road traffic simulator are the following:

Design and improvement of car equipment, where the simulator can be a powerful
tool for the on-board system evaluation. In fact, it allows to save time (and thus
money) among the different phases of a product, from the concept to the series
development, in particular between the validation with the proof of concept
and the sign off with the begin of production. In the evaluation of innovative
steering devices, the driving simulator has been used (e.g. in Toffin et al., 2003) to
investigate different torque feedbacks in the steering wheel. Results indicate that
drivers on the simulator can control their vehicles in curves with quite different
torque feedback strategies, either linear or nonlinear. However, zero torque or
inverted torque feedback makes driving almost impossible. These observations,
which cannot be implemented in real driving conditions, confirm the essential
role of coherent haptic information for driving real cars and simulators and
also suggest the existence of driver adaptation mechanisms in steering control.
Finally, simulators are used in the context of speech-controlled driver information
systems. Manstetten et al. (2001) studied how drivers interact with such a system
if it allows for natural-language communication in simulation environment. The
present contribution to the book on driver's model is more focused on this point.
The other concepts that are also considered as follows:
Training - real-time simulator is increasingly used to educate and train profes-
sional drivers (trucks, buses, etc.) and dedicated personnel (i.e. traffic control
centre) (Dois et aI., 2002).
Testing new structure - to quantify performances according to different design
options before the works start or before the commitment of resource for con-
struction.
Security and environment - with some applications in this area dealing with
studies on intelligent highways and vehicles (Sukthankar et al., 1996; 1998),
impact of road signs and infrastructure towards the driving tasks (Fitzpatrick,
2000; Horberry et al., 2004; Horberry and May, 1994; Horberry et al., 2005),
driver's behaviour analysis in critical scenarios (Pentland et al., 1999; Salvucci
et al., 2001), accident reconstruction, as well as researches on emissions and
pollution.
Studies on drivers' distraction - Karlson (2004) studied driver distraction and its
countermeasures in driving simulator evaluating the gaze direction. Lansdown
et al. (2004) investigated the impact of multiple in-vehicle information systems
on the driver, undertaken using a high fidelity driving simulator. Martens and
Winsum (2000) investigated driver's distraction in simulator using the peripheral
detection task (PDT,2 Olsson and Burns, 2000), that is a very sensitive method of
measuring peaks in workload, induced by either a critical scenario or messages

2 Peripheral detection task.

 21. Present and Future of Simulation Traffic Models 403

provided by a driver support system. The more demanding the task, the more
cues will be missed and the longer the response times to the PDT.
Rehabilitation of driving skills for persons with neurological compromise -
Rizzo et al. (2002) performed a study of 54 individuals (21 with traumatic
brain injury, 13 with stroke and 20 healthy controls) who were administered a
driving evaluation using the driving simulator (mainly based on virtual reality
technologies).
Basic research - traffic simulation is used for mathematical and statistical studies,
with the aim to improve traffic flow models, analyse interesting aspects, such as
emergent collective behaviour, swarm intelligence topics and so on (Leonardi
et al., 2004).

As next section describes, simulators can be classified according to different

factors, depending on the selected methods, on the chosen applications and on the
present constraints. In general terms and as described in Ni and Feng (2002), there
are three main actors in a simulator: traffic and environment, vehicle and driver.

21.2.2 Types of Simulator

Since the most of traffic simulator describe dynamic systems (and on the other hand
this is their most interesting application), time is always the basic independent
variable. There are two types of models: discrete and continuous. We will pay
particular attention to the discrete ones since these represent real-world systems
with enough precision and at the same time with a computational complexity
sufficiently low.
According to the level of details requested by a simulation, there are generally
three approaches in traffic simulation: macroscopic, mesoscopic and microscopic.
Macroscopic approach treats traffic as one-dimensional compressible fluid and
emphasises the general behaviour of traffic such as speed, flow, density, etc. This
means that entities, actions and interactions are described at (very) low level of
detail. Traffic stream may be represented in some aggregate manner or by scalar
values of flow rate and density. For example, individual lane changes are not
represented since the model provides global quantitative or qualitative information
(Helbing and Treiber, 1998).
The microscopic approach concerns the interactions between pairs of vehicles
and emphasises the individual behaviour of traffic such as car-following and lane-
changing; that is, most entities and their interactions are described at a high level of
detail. In this context, a lane change could invoke a car-following law with respect
to its current leader, then with respect to its putative leader and follower in the
target lane (Sukthankar et al., 1998). An example is represented by the simulator
developed in the project PELOPS (Neunzig et al., 2002; www.pelops.de). where
the traffic flow in motorways is simulated with a specific view of vehicles affair. The
basic idea is a combination of high-detailed vehicle and traffic technical models;
this permits an investigation concerning the longitudinal dynamics of vehicles
as well as the analysis of the course of traffic (traffic flow). The advantage is the
404 Tango, Montanari and Marzani

opportunity to take into account all the interactions occurring among vehicle, driver
and traffic (that is, the environment). Another practical example is called simulated
highways for intelligent vehicle algorithms (SHIVA): a tool for simulation, design
and development of tactical driving algorithms (for more details, see Sukthankar
et al. (1995)).
Finally, the mesoscopic approach lies somewhat in between. It captures some
microscopic behaviour of traffic such as car-following, but with this approach the
general characteristics of traffic (i.e. speed, flow and density, etc.) are examined.
In other words, mesoscopic model generally represents most entities at a high
level of detail, but describes their activities and interactions at a lower level. For
example, lane change manoeuvres are represented but could be performed as an
instantaneous event (Kemeny, 1993).
Another classification of simulators groups them into two sets: deterministic and
stochastic. Deterministic simulators have no random variables; all entities inter-
actions are defined through exact relationships (mathematical or logical). On the
contrary, stochastic simulators include probability functions. Deterministic mod-
els are well suited to experiments wherein scenarios are intended to be completely
reproducible. On the other hand, the vehicle behaviour may appear too mechanical
and monotonous, thus realism is somehow lessened (Champion et aI., 1999).
A relevant role in the traffic simulator is represented by the programming
methodology that should be used to create the traffic scenario. This can be pro-
grammed in a sequential way3, as for some macroscopic simulators. On the con-
trary, this can be also designed according to object-oriented principles, as for many
microscopic simulators. Starting from this approach, some relevant breakthroughs
have been achieved in the last 10 years, thanks to the development of the so-called
'distributing artificial intelligent' (Brodie and Ceri, 1992). In particular, the most
recent and widely used of these innovations is the agent-oriented method. The
traffic generator mentioned in Champion et al. (1999) has been designed using an
object-oriented method where several software agents are embedded (detailed in
the next section).
Concerning the way to compute and reproduce the driving task, three are the
main relevant levels: strategic, tactical and operational (Michon, 1985). At the
highest (strategic) level, a route is planned and goals are determined; at the inter-
mediate (tactical) level, manoeuvres are selected to achieve short-term objectives
(i.e. deciding whether to pass a blocking vehicle); eventually, at the lowest level,
these manoeuvres are translated into control operations.
Consider the typical scenario depicted in Fig. 21.1: In the figure, the host-
vehicle" (A) is in the right lane of a divided highway, approaching the chosen exit.
Unfortunately, a slow car (B) blocks the lane, preventing (A) from moving at its
preferred velocity. If A desires to overtake (pass) the slower car, a conflict can
arise with the necessity of not missing the exit. The correct decision in this case

3 That is, each program has a beginning, an execution sequence, and an end. At any given
time during the runtime of the program, there is a single point of execution.
4 With host-vehicle, it is meant the vehicle considered, i.e. the one under evaluation.
 21. Present and Future of Simulation Traffic Models 405

A c:::J Be=]-

FIGURE21.1. Typical scenario for a simulator, requiring a possible tactical decision.

depends not only on the distance to the exit, but also on the traffic configuration
in the area. Even if the distance to the exit is sufficient for a pass, there may be no
suitable gaps in the right lane ahead before the exit. Thu s tactical level rea soning
combines high-level goal s with real-time sensor con straints in an uncertain envi-
ronment. Simulation is essential in developing such systems becau se testing new
algorithms in real traffic is risky and potentially disastrou s. For example, SHIVA
not only model s the elements of the driving domain most useful to designers but
also provides tool s to rapidly prototype and test algorithms in challenging traffic
situations.
Following are some other examples: PHAROS (Reece and Shafer, 1988),
SMARTPATH (Michon, 1985) and SMARTAHS (Gollu, 1995). PHAROS focused
on important perceptu al issues and directed SHIVAs earl y development. SMART-
PATH is well suited to modelling the PATH AHS con cept (even with large numbers
of vehicles) and it uses S0I 5 animatio n package for visualisation too ls (Sukthankar
et aI., 1995). SMARTAHS is an objec t-oriented simulator, which stores its evolu-
tion state in a persistent DBMS 6 .
As discu ssed above, the most recent advances in distributed artificial intelligence
have allowed to study systems charac terised by auton om ous entities, such as a
navigation by autonomous agents for inter section management is investigated in
Dresner and Stone (2004), where multi- agent systems (MA S) is the sub-field of
artificial intelligent (AI) aiming at providing both principles for construction of
complex systems involving multiple agents and mechanism for coordination of
independent behaviour of agents. In Stone and Veloso (2000), a MAS-approach is
used to alleviate traffic con gestion, basically at intersections.

21.2.3 Case Studies of Traffic Simulator

In this sec tion, cases studies of traffic simulator are discussed. Particular attention
is paid to SCANeR II simulator? (more deta ils on SCA NeR II feature s spec-
ifications can be found at www.sca ner2.com and www.oktal.fr ). The main idea

5 Silicon Graphic, Inc.

6 Database management system.
7 SCANER(c)II is a co-branded software between OKTAL and RENAULT
406 Tango, Montanari and Marzani

. --- --- -- -- - - - -- . .
,/' Traffic

Interactively
driven vehicle

FIGURE 21 .2. SW architecture of SCANER II.

behind this simulator is that the traffic is the union of the driven vehicle and the
autonomous traffic, which interact mutually. The main design features are:

road geometry is 3D;

highway and urban scenarios are both included;
vehicle types include cars, trucks, motorcycle and bikes, pedestrians, tram,
etc ;
the development methodology is object-oriented and based on C++ language
(with the possibility of 20 and 3D visualisation).

With reference to the classifications disscused in the previous section , the model
implemented is microscopic with discrete time (possible to set, default value is
20 Hz) and stochastic. The architectural scheme is detailed in Champion et al.
(1999) and illustrated in Fig. 21.2. This architecture allows the parallel execution
of different traffic generation modules within the limits of the network capacity, if
an experiment requires a relatively high-density of traffic (few hundred vehicles) .
Each module is supported by a different machine and manages a subset of vehicles.
Regarding the simulator environment, there are two fundamental elements: the
road network and signs (RNS) and vehicle model properties, used for the de-
scription of the autonomous vehicles in the simulator. We will now discuss RNS,
whereas vehicle model properties is described in the next section.
RNS is used by the autonomous vehicles to drive on the terrain; it is defined by
a set of roads connected to each other through road nodes. Each road can have the
following features: Width, different lanes for one or two directions, road markings,
 21. Present and Future of Simulation Traffic Models 407

speed limits, road authorisations (for pedestrians, no trucks, bus only and so on),
road signs (stop, traffic lights, yield, pedestrian ways, etc.), barriers, etc.
RNS has two objects: on the road side, to order list of points, minimum two,
usually placed at the centre of an actual road; on the nodes side, this is a point placed
to connect two or more roads. Each road must be connected with two nodes: one at
the beginning and one at the ending. A road cannot be connected to the same node.
Moreover, each rode can have the following attributes: width, category, length, list
of lanes. Then, each lane can be defined by offset relative to the middle of the road,
initial and final width, direction relative to the road (same or opposite direction),
driving rules like speed limits, left or right markings, vehicle categories enabled
for the lane, etc.
In each point of a road, it is possible to place one or more signs, which refer to
a lane. Road surface database is used to compute tyre contact with the road and
consequent reactions. Usually users can set some parameters such as adherence
factor (to be used by dynamic models), spatial noise factor (how bumpy is the
road), road type (normal road, sidewalk, car park, etc.), and road nature (asphalt,
snow, concrete, etc.). Road surface and user defined information are queried by
the simulation and enter in the vehicle model to compute the behaviour of the
system, the reaction of the vehicle itself and possible feedbacks to the user. Finally,
environment objects are embedded in boxes named collision boxes used in order
to detect collision between the driving vehicle and the environment.

21.2.4 Vehicle Model Properties

As described in several research papers (Champion et al., 1999; Sukthankar et al.,
1995) all vehicles constituting the intelligent autonomous vehicles in the simulation
can be regarded, by a functional point of view, as consisting of three subsystems:
perception, cognition and control.
The architecture presented in Fig. 21.1 has the advantage to describe both
these autonomous objects and the host-vehicle in interactive way. Hence, vehi-
cles can perceive each others, use available resources (such as lanes, roads, etc.),
interact mutually and with the simulated world. This situation is illustrated in
Fig. 21.3. As a general view, this structure recalls what is stated in Panou et al.
(2005) about drivers and their modelling. There, the basic assumption is that drivers
are essentially performing a set of actions that are well known and familiar, ac-
cording to their experience. As the driving process is very dynamic, these actions
are continuously selected from a vast repository of knowledge by a diagnostic pro-
cess, which - together with the interpretation of information acquired - becomes
crucial for the dynamic sequencing of driver's activity.
In the case study considered here, the model of the driver adopted is based
on a very simple approach, which assumes that behaviour derives from a cycli-
cal sequence of four cognitive functions called PIPE: perception, interpretation,
planning and execution (Cacciabue, 2004).
408 Tango, Montanari and Marzani

Auto no mous Vehicle

I Cogn itio n I
I Act uation II+- Host-vehicle - --
1-----< Perceptio n Interactively Autonomous
Autonomous I--------e
Vehicle ~
Vehicle

\ ~
Simulated World
I \ I
I Resouces :::> Resouces
I
FIGURE 21 .3. Structure of the vehicles constituting traffic flow.

This model is not sequential; this may result from several iterations (cyclical)
of the other functions. Moreover, in agreement with the initial hypothesis, the
planning function is usually bypassed by the 'automatic' selection of familiar
frames of knowledge that are associated with procedures or sets of several actions
aiming at the fulfilment of the goal of a frame. This function is howeverimportant
as it becomes effective in unknown situations or in the case of novice drivers,
when 'simpler' frames, based on single actions or on a limited sequence of very
simple/familiar actions, are used to deal with the (new) situation.
These four cognitive functions can be associated to either sensorial or cognitive
processes and are activated according to certain rules or conditions (see Table
21.1).
Comparing the PIPE framework with the Fig. 21.3, it is possible to say that:

the first topic, perception , is common to both architecture;

interpretation and planning of PIPE are grouped in the cognition issue;
eventually, execution corresponds to the control item in the autonomous vehicle
structure.

TABLE 21 .1 . Driver's model andrules for implementation.

PIPE framework Type of proce ss Rules/governin g assumption

Perception (of signals ) Sensorial Haptic

Visual
Aural
Interpretat ion Cognitive Similarit y matching
Frequency ga mbling
Plann ing Cognitive Inference
Reasoning
Execution Behavioural Performance of selected actions
Interactions
 21. Present and Future of Simulation Traffic Models 409

In the next section, more details on the three aforementioned elements are pointed
out.

21.2.4.1 Perception Topics

Perception remains one of the most complicated tasks for humans'' and difficult
problems in implementation for simulator. In fact, control algorithms making un-
realistic perceptual assumptions can remain unimplementable on real systems. On
the other hand, perfectly modelled realistic sensors are infeasible since the simu-
lated world cannot match the complexity of real life as whole. Simulator designers
must therefore balance these alternative issues, selecting an appropriate level of
detail for their task. In the tactical driving domain, items such as occlusion, ambi-
guity and obstacle-to-lane mapping are important. Moreover, autonomous vehicle
acquires knowledge of the surrounding world in two stages: The first stage con-
sists in scanning the road ahead to detect the possible changes in driving conditions
(traffic lights and signs, pedestrian crossing, barriers, etc.), while the second stage
relates to the perception of the other vehicles.
The two simulators discussed so far (SCANeR II and the SHIVA) support
a variety of sensors and perception routines that enables cognition algorithms to
gather information about their surroundings in a realistic manner. A typical vehicle
configuration contains the following sensors:

Positioning: Dead-reckoning and/or global positioning system (GPS).

Lane tracking: Possibly with exit ramp counting features.
Car tracking modules or rangefinders.
Vehicles' self-state: Odometer, speedometer.
Potential communication system towards other equipped vehicles.

The outputs of the sensors can be corrupted by noise if desired. Some sensors are
explored in greater depth, e.g. vehicle detection in particular, since it is of critical
importance at the tactical level.
In SCANeR II, perception is based on the road network since it is used for
trajectory calculations and acquisition of information. The road network designer
should pay attention to the fact that the information has to be available to the vehi-
cles (road sections, signs positioning); moreover, two vehicles have to know each
other's status (position, speed, acceleration and direction). Autonomous vehicles
also use the knowledge of the future route of the other vehicles to foresee their
behaviour when approaching an intersection. Furthermore, designers may select

8 Drivers rely on their experience with the driving task and perceiving the road environment
relies on top-down expectations (Theeuwes, 2002). For example, Theeuwes and Hagen-
zieker (1993) have shown that drivers expect that objects that are likely to appear in a given
scene should occupy specific positions in that scene. Results from their study also showed
that errors in perceiving objects occurred when road users had wrong expectations regard-
ing the location of particular target objects. Therefore, extremely dangerous situations may
occur when the design of the traffic environment induces incorrect expectations regarding
the spatial arrangement of the object in that scene.
410 Tango, Montanari and Marzani

different types of vehicle sensors or a more abstract vehicle sensor that can be
used to prototype reasoning schemes. Once the useful information is obtained
(road curve, signs and marking, position and trajectory relative to the perceived
vehicles, etc.), a vehicle goes into the cognition phase. This phase corresponds to
the reasoning and to the decision-making processes (see also next section).

21.2.4.2 Cognition Topics

Situational awareness is the key to achieve an effective navigation in traffic; in fact,
intelligent (autonomous) vehicles should assess the outcomes of various actions
and balance the desires of higher-level goals (such as taking a specified exit) with
sensor driven constraints (observed vehicles). Moreover, these algorithms should
be tested under a variety of situations (among which the critical and dangerous
ones).
So and as aforementioned, the three cognitive processes relative to the driving
task can be characterised by decision taken at strategic, tactical and operational
levels. These are represented by a specific task-analysis, whose purpose is to
describe tasks and to characterise the fundamental features of a specific activity or
set of these. According to literature (Drury et aI., 1987; Fuller, 2005; Lansdown
et aI., 2004; Spillers, 2003), task analysis can be defined as the study of what
an operator (or a team of operators) is required to do, in terms of actions and/or
cognitive processes, in order to achieve a given objective. In this context, a goal
that a vehicle must reach is associated to each one of the aforementioned levels
and its realisation implies a choice to achieve (see Table 21.2).
For example, considering as the strategic goal the route of a vehicle driving on
a highway, it can take the next exit towards a 'service area'. In order to fulfil its
choice, the selected road, if we suppose that the vehicle is on the left lane, it has
to take the tactical decision to pass through the other lanes before the intersection.
Each choice is then implemented while being based on the lower decision level
and according to the constraints related to the vehicle's environment. So, at highest
level, autonomous vehicles have to follow their route by taking into account the
fixed and further constraints.
The tactical level corresponds to the choice of short-term objectives combining
the high level goals (route to achieve) and the constraints imposed by the lower
level (physical characteristics of the road layout, dynamics of the vehicle, traffic
flow and status). These objectives are mainly lane changes, performed in order to
optimise the route between intersections. For example, recalling again Fig. 21.1,
if vehicle (A) has to take the next exit and vehicle (B) strongly reduces its speed,

TABLE 21.2. Autonomous vehicles driving task characterised as decision levels.

Level Goal Choice

Strategic Planning a route Road

Tactical Selecting a (appropriate) manoeuvre Lane
Operational Executing the manoeuvre Speed and lateral offset
 21. Present and Future of Simulation Traffic Models 411

TABLE 21.3. Example of GMTA to a driving vehicle environment modeL

Goal/tasks Task/elementary functions involved Pre-conditions

Attain higher speed 1. Accelerate Check all conditions

(Complex task) 2. Check speedometer for elementary tasks
3. Maintain speed
Change lane 1. Check road signals - Lane free
(Elementary task) 2. Check mirrors - No incoming faster vehicles
3. Scan road side - Change lane allowed
4. Scan road forwardlbackward - Visibility
5. Set indicators (rightlleft)
6. Steer

Text in italics indicate elementary function and text in bold indicate tasks.

then vehicle (A) has to decide whether it can change lane to overtake vehicle (B)
and take exits or simply to follow (B) and wait for the same exit. It is worth to note
that tactical-level is of great importance for the feeling of realism of a simulator
and thus designers should pay particular attention to this issue.
Finally, at the operational level, if (A) decides to overtake (B), then it has to
compute the speed and the relative lateral offset appropriate with the execution of
the lane-change manoeuvre.
Always focusing on task-analysis another example is the so-called 'goals-means
task analysis' (GMTA), (see also Hollnagel (1993), where more details are pro-
vided). In this context, Table 21.3 shows a general instance for its application.
All in all, one of the major difficulties is the hand-tuning of the reason-
ing/deciding object parameters. Thereby, a careful investigation of learning tech-
niques is one of the main activities to carry out in order to make this task automated.
Potential solutions are described in Baluja (1994).

21.2.4.3 Actuation/Control Topics

This part provides the actuation of what has been decided by the cognition sub-
system, the operational level. Acceleration and wheel angle are computed so that
the autonomous vehicles can move. Then, vehicles can get also the status of other
objects in the environment and thus performing their task while trying to avoid
accidents (unless they have to be simulated, i.e. due to a high driver's impairment
level).

21.2.4.4 Implementation of Vehicle Model

Several kinds of vehicles can be represented, such as cars, tractors, pedestrian,
motorcycle, etc., each one with a specific set of parameters (e.g. type, dimensions,
weight, including loaded and unloaded status, position and length of axles, driver's
eye position, wheel radius and tyre categories, engine category, maximum speed,
wheel angle amplitude, acceleration, brake and inertia factor). The car model used
412 Tango. Montanari and Marzani

Aerodynamic
Forces

Comfort
(vibration due to road
surface)

Road
DB

FI GURE 21.4. Scheme of sub-models for the dynamic vehicle module.

is a simplified dynamic model , constituting of a set of sub-models, as illustrated

in Fig. 21.4 .
Concerning the dynamic beha viour, the various model s involved in the simu-
lation s are elaborated in real-time with a defined refresh-time ( 100 Hz), which is
often called simulation step. At every simulation step the simulation considers: Ve-
hicle co nfiguration as defined by the driving scenario; all the kine matics variables;
external input s such tho se co ming from the driver or environmental conditions;
eventually, type and topo graph y of the road under the tyres, as defined by the
SCANeR II RNS .

21.2.5 Two Examples ofApplications with Traffic Simulator

In this paper, two exampl es of applications of traffic simulator are illustrated. One
is surveyed by literature and in particular it is derived by an activity carried out
at University of Michigan, in which a study is performed about the evaluation of
adaptive cruise control (ACC) and colli sion warning (CW) using a longitudinal
driver's model. In this context, onl y an overview is given on the micro scopic traffic
simulator used, but more details can be found in Lee (2004) .
The other example is a case of a simulator which is recentl y installed within
a laboratory on Mechatroni cs named MECTRON and placed in Reggio Emilia
(North of Italy)", The simulator is used for human factor s and human machine

9 The MECTRO N Laboratory (www.mectron.org) re-enters within the projects of regional

laboratories of the district for the HIgh MECHanical technology (HI-MECH). The project
is financed through the regional program for the industrial research, the innovation and
the technology transfer (PRRIITT) of the Emilia-Romagna region. The goal of Mectron is
 21. Present and Future of Simulation Traffic Models 413

interfaces studies, including both innovative devices for primary and secondary
task. The simulator and the corresponding research activities are carried out by the
Human Machine Interaction Group of University of Modena and Reggio Emilia 10

21.2.5.1 The University of Michigan Microscopic Traffic Simulator

A microscopic simulator, developed at University of Michigan and called UM-
ACCSIM, was used to evaluate the performance of ACC and CW systems. This tool
has the possibility to simulate and record the motion of each vehicle operating on a
two lane circular track (as in SHIVA simulator) and to produce many microscopic
and macroscopic outputs. Because there are two lanes, lane change behaviour is
one of the major source of speed variation for the leading vehicle; therefore, the
implemented lane change algorithms take into account both the intention and the
safety factors. The first one is the period of time during which the host-vehicle is
driven below its desired speed due to the slower lead-vehicle; when it becomes
higher than a certain threshold, the host-vehicle decides to make a lane-change
(this choice should depend on the driver's model implemented). For the second,
the host-vehicle looks at the nearest adjacent lane vehicles, front and rear, in order
to determine if a lane-change can be safely executed. When both these factors are
satisfied, a lane change manoeuvre occurs.
Concerning the driver's model that has been used, two main ones have been
taken into account (for more details, see Lee (2004)): The first is by Liang and
Peng (2000), where the dynamic speed control behaviour of human-driven vehicles
is represented, but where a linear-follow-the-leader model is used, with the same
parameters for all vehicles; the second model is the modified Gipps model, where
the model itself and the parameters are identified from real human driving data
and thus all aspects of longitudinal human driver's model in controlling vehicle
motions are based on statistical driver's behaviour data (for more details, see Gipps
(1981)). Preliminary results showed that the best one (fittest and more robust) was
the second and thus the Gipps model was used and since human driving behaviour
is unstable and this model with its set of parameters is not guaranteed to be stable
either, emergency collision avoidance algorithms were implemented. Here it is
worth to note that, if all behavioural models describe human-controlled vehicles
with enough fidelity, the whole simulator can be expected to demonstrate many
characteristics exhibited by human drivers.
When the traffic density is low, no shock wave is observed in the simulations
and lane changes can occur. When the traffic density is high, lane changes become
rare, since safe empty spots available for such a manoeuvre are harder to access.
In this sense, the behaviour is more similar to the one of a single lane. Initially, no
wave propagation behaviour is observed, but because of the fact that all vehicles

to design, prototype, test and develop innovative solutions for mechatronic systems, such
as control by-wire applications, human factors for mechatronics' systems (e.g. Ergonomic
by-wire steering-wheels), the study of fluid power for mechatronics and advanced materials.
10 www.hmi.unimore.it.
414 Tango, Montanari and Marzani

have their own desired speed, some vehicles move closer and thus begin to interact
with each other. As this type of interactions increases, deceleration and accelera-
tion waves are propagated upstream. On the other hand, if traffic is string-stable,
these waves would be attenuated. Therefore, traffic density is expected to be a
fundamental factor in wave propagation velocity, with average speed that mono-
tonically decreases as density increases. In high-density case, emergency braking
algorithms are used with much greater frequency.
In this section only a general and brief overview of the main results achieved
by a microscopic traffic simulator have been presented, in order to evaluate the
characteristics of the longitudinal driver's model. More details on this research are
available in Lee (2004). All in all, a modified Gipps model has been used to describe
the longitudinal human driving behaviour (and so the' personality' of each vehicle-
entity in the simulator). Traffic flow simulations for several traffic densities were
carried out: the resulting traffic flow and the average velocity results agree with real
traffic data. This means that the modified Gipps model and all the related parameters
can represent in a satisfactory way the macroscopic traffic characteristics, such as
traffic flow rate and average velocity with respect to traffic density. The impact
of intelligent transportation systems in this context is expected to be high and to
influence the traffic rate as well as the speed of shockwave propagation.

21.2.5.2 The MECTRON-HMI Group at University of Modena

and Reggio Emilia Driving Simulator used in Human factors and Human -
Machine Interfaces Studies.

This section presents a research carried out by the Human Factors Laboratory of
the MECTRON project (details on MECTRON are reported above). As previously
introduced, the Human Factors Laboratory is led by the HMI Group of University
of Modena and Reggio Emilia.
The main objective of this research is to develop specifications and requirements
of a steer-by-wire device, which has the peculiarity to give primary importance
to the human factors and ergonomics aspects (a so-called ergonomic steer-by-
wire, i.e. ESBW). In particular, this is illustrated and investigated in terms of
driving quality and performances, vehicle controllability 11 and human cognitive
functionalities (such as mental workload and situational awareness). Here, only
the experimental set-up concerning the use of traffic and environment simulation
is pointed out, where the main scenarios are described.
At the moment, the experiments are in course. The basic idea of this work is that
a by-wire device is able to support properly the driving task only if the drivers'
perception in the usage of this system remains the same of a traditional steering
wheel.

11This concept is used as within the ED Integrated Project PREVENT (and particularly it
sub-project RESPONSE 3). For more information, see the web site www.prevent-ip.org/
en/prevent_subprojects/horizontal_activities/response_3 .
 21 . Presentand Future of Simulation Traffic Models 415

FIGURE 21 .5. Drivingscenarioin OKTAL simulator.

In this context, the simulated environment is based on SCANeR II and it

presents the following characteristics: operative scenarios for the single external
scene (highways, rural roads, etc .) combined with simple manoeuvres (such as
rural roads plus lane change, highway plus lane keeping, etc.). An example of this
driving scenario is illustrated in Fig . 21.5 . The simulator allows joining different
technologies and devices, in order to:

recon struct the driving scenario (route , weather conditions, traffic density) in a
reliable way and in real-time;
project the scenarios on the dedicated screen ;
perform (and promote) the interaction between user and system, by means of a
reproduction of the automobile cockpit, the use of steering wheel, accelerator
and brake pedals, with force feedbacks and feed forward (to reproduce the same
feeling and dynamics of a car) .

In the aforementioned experiment by this tool, the rules describing the behaviour
of the force-feedback and force-feed forward actuator in the ESBW are under
investigation. To perform such analysis, a previous benchmark on the different
types of steering devices and reactive torque algorithms was evaluated. The analysis
focused on two different kinds of force-feedback; one concerning reactive torque
algorithms based on vehicle behaviour (e.g. lateral acceleration, yaw rate, sideslip
angle) , the other one based only on steering-wheel dynamic behaviour (e.g. steering
angle, steering rate) . In literature, both were investigated from a human factor
point of view, that is the results of tests of driver performances in different driving
scenarios (e.g. lane change, car following, highway, rural road) under primary
and/or secondary tasks . The comparison among these results allowed to define the
best user-centred algorithms, then to design different solutions of an ergonomic
steer-by-wire featuring several force-feedback models . Finally, these prototypes
416 Tango, Montanari and Marzani

are in course of implementation on the steering wheel of the driving simulator; the
aim is to evaluate, during driving tests, the most performing algorithm in terms
of driver response. These assessments are performed comparing the following
dependant variables:

Driver's performances in driving task.

Mental workload, situational assessment and safety level of the user in executing
the primary and secondary tasks using both subjective scale, as rating scale
mental effort (RSME 12, (Zijlstra, 1993)) and objective evaluation, as peripheral
detection task (PDT).
Subjective evaluation of the driving device, on the basis of scale of preference
and driving quality as driver opinion scale (DOS) and driving quality scale (DQS;
Nilsson (1995).
Factor analysis of driver's distraction in well-defined driving contexts/scenarios.
Error analysis on determined scenarios, due to the introduction of new actuators
and devices.

At the moment, experiments are in progress and it is not possible to anticipate any
results. Nevertheless, the simulator will be used not only as an assessment tool but
also to observe how the ESBW can become a supportive system aimed at improving
drivers' situation awareness in specific conditions (e.g. via the implementation of
proactive supportive actions as a major resistance of the steering-wheel in case
the drivers is trying to change the lane, risking a collision with a vehicles which
is already occupying the adjacent lane). These studies will be carried out in the
late phase of MECTRON project, once the ESBW assessment will be properly
completed.
Traffic simulators aim at investigating the impact of these adaptive interfaces 13
with reference to drivers' reactions and expectations; this can represent a valuable
perspective for these tools. Nevertheless, the only simulation, based on traffic,
could not be enough: thus a mutual interaction with environment, drivers' models
and vehicles is needed, as described in details in the next section (introducing
another relevant case study of AIDE project) and regained in the conclusions.

12 Rating scale mental effort.

13 According to Hoedemaeker et al., 'by an adaptive support it is meant a system that in
some way takes into account the momentary state of the driver, in particular his present
level of workload, in determining the appropriate timing and the content of the supporting
message or intervening activity the system will produce. This should thus prevent the driver
from becoming overloaded or distracted because of the impending multitude of information.
These are based on an analysis of potential bottlenecks, including risk estimates derived
from the prevailing traffic situation, so that it can be determined what types of support
messages are allowed in a particular situation'. This involves both primary (as an adaptive
steering wheel able to warn the driver in case of potential collision) and secondary tasks (as
vehicular phone which stopped an incoming call in case of critical driving condition).
 21. Present and Future of Simulation Traffic Models 417

Speed
Acceleration
Yaw-Rate
Deviation

FIGURE 21 .6. Closed-loop schemefor DYEsimulation.

21.2.6 Integration of Driver, Vehicle and Environment

in a Closed-Loop System: The AIDE Project
The European Integrated Project AIDE 14 has been explicitly planned to develop
concepts for safe and efficient human-machine interaction as a key step towards
the realisation of intelligent road vehicles with higher safety and value-added
services, as pointed out in the e-safet y report (European Commission, 2002 ). In
particular, one of the main sub-goals of the AIDE integrated project is to perform
theoretical anal yses and field studies aimed at predi cting human behaviour and
error, specifically related to the interaction with advanced driver assistant systems
(ADAS) and in-vehicle inform ation systems (IYIS ).
In order to achie ve these goals, the simulation model should include a mutual
interaction among the driver, the vehicle and the environment. Thi s is a precise
approach named driver- vehicle-environment (DYE) and the corre sponding model
and simulation is expected to retain the essential correlations between the funda-
mental independent variables, predicting the driver's behaviour in dynamic and
rapidly changing conditions. Therefore, the DYE can be seen as closed-loop sys-
tem as shown in Fig. 21.6 .
Such a system can work as follow s: the intell igent driver sub-model collects
information related to its driving task. The information come s from three source s:
Environment stimuli, vehicle feedbacks and driver characteristics. Some pieces of
the information may need to be ' fuzzyfied'. A possibility is to use a neural network,
which is capable of associating input patterns with their corresponding output
patterns, ju st as human being s have the ability to deal with unknown situation,
based on his/her knowledge and past experience. According to the information
fed in, the neural network suggests the corre sponding control strategies, i.e. gas,
brake and steering angle (more detail s on this point are reported in the conclusive
discussion).

14 AIDE stand adaptive integrated driver- vehicle interface - IST-I-507674-IP).

418 Tango. Montanari and Marzani

~
FIG URE 21.7. The SHELL configur-
ation.

E
I

L L H
\
l~
S
S Softwa re
H Hardware
E Enviro nme nt
L Liveware

21.2.6.1 General DVE Architecture

Inside the AIDE project (and particularly in the sub-project which is taking care
of the simulation related topics, i.e. the sub-project 115 ) , the development of the
architecture adopted to represent the interactions between driver, vehicle and en-
vironment is framed in a generic architecture describing the way in which human s
interact with the world and systems around them (Panou et al., 2005). This is
called SHELL (Edwards, 198 8), whose structure describe s the connections exist-
ing between humans , defined Liveware (L), and the other elements of the work-
ing environments (driving, in this case), as described in the following list and in
Fig. 21.7:

I. The physical source s, such as equipment, systems , interface s and machines ,

which may be termed Hardware (H): Corresponding interaction between Hard-
ware and Liveware are mentioned as L-H interactions.
2. Rules, regulations, laws, procedures, customs, practices and habits governing
the way in which a plant or a machine are operated, called Software (S): L-S
interactions.
3. Socia l, physical and technical aspect s of working contexts, which may be called
Environment (E): L-E interactions.
4. Direct comm unications and exchange of information of the driver with other
human beings , such as passengers or other drivers in different vehicle s, which
is termed Liveware (L): L-L interactions .

The model focu ses on the way in which a process of interaction is influenced
and can be simulated with respect to a single journey. The model intends to deal

15 More details in www.aide-eu .org,

 21. Present and Future of Simulation Traffic Models 419

with dynamic and adaptive characteristics of a driver, driving environment and

vehicle; in principle, it can be continuously updated and modified.
Hence, the SHELL architecture can be used to represent the different variables
that affect the DVE model. Moreover, in order to capture the crucial dimension
time, affecting the process of driving in different conditions and situations, a fur-
ther sub-division can be considered; in particular, three discrete time levels are
envisaged:

Static variables, which account for variables that do not change over the journey
(examples of these are age of driver, gender, personality, procedures, etc.).
Quasi-Static variables, which account for the phenomena and interactions that
may change during ajourney even though these changes are slow and foreseeable
(examples of quasi-static variables are attitudes, behavioural adaptation, etc.).
Dynamic variables, which account for the events and phenomena that occur
during a journey and may be affected by the evolution of the DVE interaction
itself or may not be anticipated (examples of dynamic variables are workload,
stress, traffic conditions, type of roads, weather conditions, traffic situations,
etc.).

The next section details the time-frame concept.

21.2.6.2 Time Frame for DVE Model

The time distribution of the model takes into consideration a single time-step
prediction system which is continuously updated by real data, resulting for the
actual behaviour of the overall DVE environment. Thereby, for each time step of
the simulation (t == ti), the overall contribution of the three components of the DVE
model is evaluated by extracting the variables that influence an index namely driver
impairment level (DIL). This index is evaluated on the basis of several variables,
both subjective (risk-taking, complacency, etc.) and objective (such as risk-level
and performance limitations), as well as their linear combination (for more details,
see Panou et al., 2005). Moreover, the error generation mechanism is related to the
DIL index, to the dynamic interaction process of the three elements of the DVE
model and to the 'failure mode and effect analysis' (FMEA) process carried out by
the component, which manages the DVE simulation. So, the simulation manager
firstly performs the FMEA and identifies whether the driver is able to continue
the ongoing activity, that is DIL is over a defined threshold (0 ::s DI L(t == ti) ::s
DI L 1) or if driver performs inadequately or if he/she is impaired (D I L > DILl).
In particular, the effect analysis selects the actual effect that should be shaping the
activity in the following time step. Fig. 21.8 shows the dynamic DVE architecture
in AIDE.
Such a process of FMEA discussed for the Diver model can be replicated for the
models of the vehicle and environment in the case of a full simulation of the DVE
interaction, thus generating possible inadequate or improper performances of the
vehicle or risky situation associated with the environment. In a real situation the
420 Tango, Montanari and Marzani

DRI VER I I = I, I VE HICLE

~
~lale OfD (IJ SlaleOf~7IJ/
.J 1=1,+1'11

1= 1,+1'11 \ ( par.AjJ. V al 1,+1'11

Par.Aff. D all, +1'11
Simulation Manager
DIL(I=I;+I'1I)=f (. .)=> Fail. Mode
Effe ct Analysis => Ca ll . offai lure

EN VIRON M ENT

FiGURE 21.8. Architecture of time-evolution DYEmodel.

dynamic evolution of both vehicle and environment can be recorded and compared
to the simulated data (or derived by them) .
All in all, the activity on DVE model simulator is still in progress and therefore
current formulation may be not cover a number of variables and cases which then
could become very important. This will be clarified by the experiments envisaged
inside the project AIDE, particular in the topic-related subproject.
However, the overview provided in this section is enough to describe the main
idea and objective of the DVE simulator and the rationale of a promising perspec-
tive for the traffic simulator: An integration of the three main actors involved in
the driving scenario that is the driver, the vehicle and the environment. In this case,
the main topic is not the development (at least, not only) of a specific algorithm
for the simulation of traffic/environment or the implementation of a new driver's
model, but to point out the interaction of these three elements, in an integrated
approach.

21.3 Conclusions and Further Steps

In simulator experiments, different needs and different situations must be consid-
ered, as well as specific conditions have to be reproduced, such as insertion on
highways, accidents, reduction of the number of lanes, etc. In these cases, con-
flicts between vehicles may appear and the flow stops even if in real world such
conditions would not cause congestions. In order to deal with this trouble, activ-
ity carried out in distributed artificial intelligent (DAI) and multi-agent systems
(MAS) are taken into consideration for the traffic simulation. We will now into
details about these advanced works .
 21. Present and Future of Simulation Traffic Models 421

21.3.1 Towards a Multi-Agent Approach

As aforementioned, DAI is the area of computer science coping with a set of en-
tities aimed at imitating the human intelligent behaviour. Road traffic belongs to
this field, as described in Ossowski et al. (2004). A MAS is a set of software which
coordinate beliefs, desires and intentions, knowledge, goals and plans of human
entities, so as to act or solve problems, including the coordination among agents
themselves. To be regarded as 'intelligent' , these entities have to show behaviours
that are rational, autonomous and capable of communication and actions. In the
real environment, a vehicle and its driver is an agent; in simulator, an autonomous
vehicle (including the driver as well) and an 'inter-actively driven vehicle' are
agents. Anyway, both in real and in simulated situations, the traffic is constituted
of multi-agent systems and entities. In case this view is missing-due to an ab-
sence of communication, for example-an individualistic behaviour appears. The
consequence is a penalty of a part or of the whole traffic stream; vehicles, in fact,
are able to move but not to coordinate their knowledge and goals to get deal with
some of the illustrated drawbacks.

21.3.2 New Developments and Prospective

New possible developments concern each of the three main actors described in our
approach, that is the driver, the vehicle and the environment, presented as follow-
ing. The general statement is that traffic is composed by autonomous entity with
(basically) the capacity of perceiving their local environment and communicating
each other.
The first entity is to model the driver-vehicle control-loop system. Starting from
the vehicle, a description of possible configuration and architecture has been pro-
vided; this is of course one of the possible different approaches. For example,
if a four-wheels vehicle model is considered rather a two-wheels, this can vary
the realism of object behaviour and thus of the simulation. Furthermore, by a
physical point of view, 2D dynamics accuracy might not impair real-time perfor-
mances, such as engine and gearbox. The other fundamental aspect is the drivers
and their behavioural and cognitive model (i.e. psychological security distance,
average latency at cross-sections and so on). In this sense, one of the more ad-
vanced projects dealing with these topics is still AIDE, as discussed earlier, where
we have addressed a specific model of human behaviour enabling to predict ac-
tions and sufficiently adaptive to individual characteristics, in order to personalise
the interface between driver and technological system. For example, based upon
data gathered over long periods of driving and under different conditions, driver's
reaction time is estimated and clustered into four levels: very slow, (i.e. above 1.3
s), slow (above 1 s), average (0.8-1 s) and quick (below 0.8 s). Driver's average
lane position and change (based upon time to line crossing (TLC) measurements)
are also estimated. Other preferences, such as provision of navigation and route
guidance by map, will be estimated by monitoring the driver's own selections
through a series of system usages.
422 Tango, Montanari and Marzani

All these three items can be handled by two different agents. Initially, a user's
profile configuration agent will support different 'types of users' , with some pref-
erences selected by the users themselves. Then, a customisation agent will monitor
the user's driving behaviour and preferences/actions, by keeping and processing
the user's driving record, i.e. average position in the lane, average headway, typical
speeding and braking pattern, preferred seating position, average use of radio and
mobile phone, other services, like navigation, requested often, etc. The self-built
user profile will be always possible to be reviewed and changed by the user.
The Java agent development framework (JADE) is the basis for building inter-
operable agents FIPA 16 compliant. Such a framework ensures that agents' aspects
(message transport, encoding, parsing, agent life cycle, etc.) are dealt. Data will be
stored in a driver smart card (for example), in order to be used in other equipped
cars. This personalisation can help to better estimate the DIL, because it considers
also the effects due to the differences between different driver's typology (recalling
the PIPE framework, at the execution level, as pointed out by the AIDE project).
Furthermore, also taking into account the effects produced by the use of ADAS
and IVIS applications can be included in the human model (i.e. in terms of be-
havioural adaptation), thus enlarging the applicative scenarios of use of the simu-
lation.
All these aspects can contribute to make easier and more realistic the creation
and implementation of driver-vehicle systems as autonomous agents, in order to
deal with a wide variety of traffic conditions and applicative scenarios.
The other item to be considered is the environment, which is closely related to
the local perception in simulated world. We have already described the so-called
RNS framework, where the road network has to be apprehended according to the
local environment of the agent (vehicles ahead, road type, road layout, etc.). In
order to fulfil this perception issue, it is necessary to take into consideration the
interaction between the driven vehicle (that is, the vehicle with more sophisticated
human model and specific system on-board) and the other autonomous agents. The
study and the implementation of the environment model is one of the main tasks
for a traffic simulator, since it could not be too different from real scenarios, but on
the other side a complete imitation of real world is not possible. Therefore several
approaches are used, both considering particular aspects of the traffic management
(as in Dresner et al. (2004) and Qi (1997)), investigating specific types of on-board
systems as collision warning, adaptive cruise control, etc. (Christen and Neunzig,
2002; Miller and Huang, 2002; Ozguner et al., 2004). The ambitious goal of joint-
DVE approach is point towards the merging of these works, including the impact
of ADAS/IVIS applications with a traffic model.

21.3.3 Open Points and Future Steps

In all simulators the main goal is to achieve a realistic behaviour of each entity,
that is DVE in its design and its evaluation. The basic idea pointed out in this

16 Foundation for intelligent physical agent.

 21. Present and Future of Simulation Traffic Models 423

paper has been about the perception issue: even if data are sometimes insufficient
(as quantity) or erroneous (as in real life), a perception based on (simulated)
visual information can alone enable the vehicles to communicate each other their
objectives. For example, if a simulated scenario involves a motorway network,
an insertion has to be simulated (maybe with a congestion caused by an accident
occurred downstream) and vehicles should be able to perceive surrounding entities
intentions and to accord with (for example, entering the motorway, changing lane
manoeuvre, etc.). In other words, it is necessary to implement a structure that
enables an agent to apprehend and answer any request concerning a resource
access: If it is common and shared, agents have to cooperate in order to coordinate
their actions. In case vehicles do not have this capacity and even if they observe the
highway code, a vehicle wishing to access the highway does not have any priority
and thus it risks staying stuck on the access-ramp.
Many architectures nowadays do not allow a cooperative behaviour immedi-
ately. As pointed out, several aspects have to be considered and merged together,
concerning the perception, the cognition and the interaction. They need to be im-
plemented following a DAI/MAS approach. Whilst the agents have only a reactive
behaviour, it is possible to note a multi -agents coordination: in the example pre-
viously shown, flow of vehicle is strongly slowed down, which is normal, but it is
not stopped. In other words, it can be said that an emergent behaviour arises and
different research works can be carried out on this item.
To sum up, considering traffic as a multi-agent system allows to simulate a more
realistic but simulated environment, with which also (more) complex scenarios can
be analysed, by starting a coordination of traffic entities. This environment model
can be coherent with a homogenous integration of the interactively driven vehicles,
using a joint approach of user's behavioural and personalisation model (including
the presence of driving supporting systems on-board). This is the actual area of
research and work of some projects, such as AIDE. In this context, the environment
model is built by considering driving behavior from the point of view of how drivers
perceive, attend and memorize environmental conditions to make choices and take
proper actions to those conditions.
One of the possible future activities is to model how these conditions are related
to risk factors, in order to determine which are the most critical scenarios. There-
fore, the DVE model should include all those parameters from the environment
which drivers indicate as the most attention demanding. This type of environment
model is synthesised into a preliminary joint DVE model, as illustrated before.
Another one, and related to the previous by a certain viewpoint, concerns the
evaluation and assessment of predictive models of driver's workload, in different
contexts and using specific hap-tic devices (such as the steer-by-wire prototype,
as aforementioned).
It is worth to add, however, that a DASIMAS approach cannot be regarded as
a panacea for each problem in simulation, because drawbacks and shortcomings
exist in this case as well; in fact, possible criticism could be related to the use of
neural networks performance as a decision-making framework, since sometimes
revealed limits when applied in real-time systems due to the time spent in learning
phase if any and/or in the evaluation of proper solution among a list of alternatives.
424 Tango, Montanari and Marzani

Therefore, the activities on these topics are continuously running. Some sugges-
tions to overcome possible difficulties of tuning the reasoning object parameters
concern the investigation of learning techniques, such as in Baluja (1994), where
an abstraction of the basic genetic algorithm, the equilibrium genetic algorithm
(EGA), is reconsidered within the framework of competitive learning. This paper
explores population-based incremental learning (PBIL), a method of combining
the mechanisms of a generational genetic algorithm with simple competitive learn-
ing. Moreover, an empirical analysis is presented, with a description of a class of
problems in which a genetic algorithm approach may be able to perform better.
Among the cases discussed in this document, the DVE simulator could be a possi-
ble field of application, especially the part concerning the traffic and environment
agents, based on a training data set constituted by real road data from several
driving session performed by different drivers in different scenarios.

References
Balaban, C.D., Cohn, r.v, Redfern, M.S., Prinkey, r, Stripling, R. and Hoffer, M. (2004).
Postural control as a probe for cognitive state: Exploiting human information processing
to enhance performance. International Journal of Human-Computer Interaction, 17,
275-286.
Baluja, S. (1994). Population-based incremental learning: A method for integrating genetic
search based function optimisation and competitive learning. Technical Report CMU-
CS-94-163. Cornegie Mellow University, Pittsburgh, PA, USA.
Bertacchini, A., Minin, L., Pavan, P. and Montanari, R. (2006). Ergonomic steer by wire:
Caratteristiche e modalita funzionale. Deliverable R5.1, MECTRON Laboratory, Reggio
Emilia, Italy.
Brodie, M.L. and Ceri, S. (1992). On intelligent and cooperative information systems: A
workshop summary. International Journal of Intelligent and Cooperative Information
Systems, 1(2),249-289.
Cacciabue, P.C. (2004). Guide to Applying Human Factors Methods-Reference Model of
Cognition Chapter, Springer, Berlin.
Champion, A., Mandiau, R., Kolski, C., Heidet, A. and Kemeny, A. (1999). Traffic genera-
tion with the SCANeR II simulator: Towards a multi-agent architecture. In Proceedings
of DSC'99, Paris, France, pp. 311-324.
Christen, F. and Neunzig, D. (2002). Analysis of ACC and stop and go using the simulation
tool PELOPS. Final Technical Report IKA.
Dixon, K.R., Lippitt, C.E. and Forsythe, lC. (2005). Modelling human recognition of
vehicle-driving situations as a supervised machine learning task. In Proceedings of the
11th Conference on Human-Computer Interaction.
Dols, r, Pardo, r, Breker, S., Arno, P., Bekiaris, E., Ruspa, C. and Francone, N. (2002). The
TRAINER project: Experimental validation of a simulator for driver training. In Driving
Simulation Conference, Paris, France.
Dresner, K. and Stone, P. (2004). Multiagent traffic management: A reservation-based inter-
section control mechanism. In The Third International Joint Conference on Autonomous
Agents and Multiagent Systems, New York, USA.
Drury, C.G., Paramore, B., Van Cott, H.P., Grey, S.M. and Corlett, E.N. (1987). Task
analysis. In G. Salvendy (Ed.). Handbook of Human Factors. J Wiley, New York, pp.
370-401.
 21. Present and Future of Simulation Traffic Models 425

Edwards, E. (1988). Introductory overview. In E.L. Wiener, and D.C. Nagel (Eds.). Human
Factors in Aviation. Academic Press, San Diego, CA, pp. 3-25.
European Commission (2002). Final report of the esafety working group on road safety.
Information Society Technologies, Bruxelles.
Ferrarini, C. (2005). Analisi comparata dell'usabilita di un software per la supervisione e
il controllo dei processi industriali: il caso Moviconx. MS Thesis, University of Modena
and Reggio Emilia.
Fitzpatrick, K. (2000). Alternative design consistency rating methods for two-lane rural
highways. FHWA Report RD-99-172.
Fuller, R. (2005). Towards a general theory of driver behavior. Accident Analysis and
Prevention, 37(3),461-472.
Gipps, P.G. (1981). A behavioural car-following model for computer simulation. Trans-
portation Research, 15B, 105-111.
Gollu, A. (1995). Object Management Systems. PhD Thesis.
Helbing, D. and Treiber, M. (1998). Jams, Waves and Clusters. Science, 282, 200-201.
Hoedemaeker, M., De Ridder, S. and Jahnsenn, W. (2002). Review of European human
factors research on adaptive interface technologies for automobiles. TNO Report, TM-
02-C031.
Hollnagel, E. (1993). Human Reliability Analysis: Context and Control. Academic Press,
London.
Horberry, T., Anderson, 1., Regan, M. and Tomasevic, N. (2004). A driving simulator
evaluation of enhanced road markings. In Proceedings of the Road Safety Research,
Policing and Education Conference, Perth, Australia. Peer reviewed paper.
Horberry, T. and May, J. (1994). SpaD Human Factors Study: Directory ofRelevant Liter-
ature. Applied Vision Research Unit. University of Derby. UK
Horberry, T., Regan, M. and Anderson, 1. (2005). The possible safety benefits of
enhanced road markings: A driving simulator evaluation. Transportation Research
Part F, 9, 77-87.
Kangwon, LJ. (2004). Longitudinal Driver Model and Collision Warning and Avoidance
Algorithms Based on Human Driving Database. PhD Thesis. UMI 2004.
Karlson, R. (2004). Evaluating driver distraction countermeasures. Master Thesis in Cog-
nitive Science, Cognitive Science Study Program, Linkopings universitet, Sweden.
Kemeny A (1993) A cooperative driving simulator. In Proceedings of the International
Training Equipment Conference and Exhibition Conference (ITEC), London, pp. 67-71.
Lansdown, T.C., Brook-Carter, N. and Kersloot, T. (2004). Distraction from multiple in-
vehicle secondary tasks: Vehicle performance and mental workload implications. Er-
gonomics, 47, 91-104.
Leonardi, L., Mamei, M. and Zambonelli, F. (2004). Co-fields: Towards a unifying model for
swarm intelligence. In Engineering Societies in the Agents World III: Third International
Workshop (Vol. 2577, pp. 68-81). Lecture Notes in Artificial Intelligence. Springer-
Verlag, Berlin.
Liang, C.Y. and Peng, H. (2000). String stability analysis of adaptive cruise controlled
vehicles. JSME International Journal Series C. 43(3), 671-677.
Lieberman, E. and Rathi, A.K. (1997). Traffic simulation. InN.H. Gartner, CJ. Messer and
A. Rathi (Eds.). Traffic Flow Theory (chapter 10). Federal Highway Administration and
Oak Ridge National Laboratory.
Manstetten, D., Krautter, W., Grothkopp, B., Steffens, F. and Geutner, P. (2001). Using
a driving simulator to perform a wizard-of-oz experiment on speech-controlled driver
information systems. In Proceedings of the Human Centered Transportation Simulation
Conference (HCTSC 2001), Iowa City.
426 Tango, Montanari and Marzani

Martens, M.H. and van Winsum, W. (2000). Martens, M. H. and van Winsum, W.
(1999). Measuring distraction: The peripheral detection task. Online paper. Available at
www.nrd.nhtsa.dot..gov/departments/nrd-13/driver-distractionlwelcome.htm.
Michon, J .A. (1985). A critical view of driver behaviour models: What do we know, what
should we do? In L. Evans and R. Schwing (Eds.). Human Behaviour and Traffic Safety.
Plenum Press, New York, pp. 485-520.
Miller, R. and Huang, Q. (2002). An adaptive peer-to-peer collision warning systems. IEEE
Vehicular Technology Conference (VTC). Birmingham, AL, USA.
Minin, L. (2005). Prototipazione virtuale nell'interazione uomo-macchina: Benchmark
analysis e testcomparati per supportare la scelta dei software di prototipazione virtuale
piu adeguati a differenti tipologie di interfacce utente. MS Thesis, University of Modena
and Reggio Emilia.
Neunzig, D., Breuer, K. and Ehmanns, D. (2002). Traffic and vehicle technologies. Assess-
ment with simulator PELOPS. ISSE Conference.
Ni, D. and Feng, C. (2002). A two-dimensional traffic simulation model. In Proceedings of
the 7th Driving Simulation Conference.
Nilsson, L. (1995). Safety effects of adaptive cruise controls in critical traffic situations. In
Proceedings of The Second World Congress on Intelligent Transport Systems. pp. 1254-
1259.
Olsson, S. and Bums, P. (2000). Measuring distraction with a peripheral detection
task. The National Highway Traffic Safety Administration Driver Distraction Inter-
net Forum, on-line paper, available at http://www-nrd.nhtsa.dot.gov/departments/nrd-
13/driver-distractionlPapers20006.htm#A6
Ossowski, S., Fernandez, A., Serrano, J.M., Perez-de-la-Cruz, J.L., Belmonte, M.V.,
Hernandez, J.Z., Garcia-Serrano, A. and Maseda, J.M. (2004). Designing multiagent
decision support system-The case of transportation management. Third International
Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS 2004), New
York,pp.1470-1471.
Ozguner, F., Ozguner, U., Takeshita, 0., Redmill' K., Liu, Y., Korkmaz, G., Dogan, A.,
Tokuda, K., Nakabayashi, S. and Shimizu, T. (2004). A simulation study of an inter-
section collision warning system. In Proceedings of the International workshop on ITS
Telecommunications, Singapore.
Panou, M., Cacciabue, N., Cacciabue, P.C. and Bekiaris, A. (2005). From driver modelling
to human machine interface personalisation. IFAC 16th World Congress, Prague, Czech
Republic.
Pentland, A. and Liu, A. (1999). Modelling and prediction of human behavior. Neural
Computation, 11, 229-242.
Qi, Y. (1997). A Simulation Laboratory for Evaluation of Dynamic Traffic Manage-
ment Systems. PhD Thesis at Massachusetts Institute of Technology, Boston, MA,
USA.
Recarte, M.A. and Nunes, L.M. (2003). Mental workload while driving: Effects on vi-
sual search, discrimination, and decision making. Journal of Experimental Psychology:
Applied, 9, 119-137.
Reece, D.A. and Shafer, S.A. (1988). An overview of the PHAROS traffic simulator. In
Rothengater, J .A. and De Bruin, R.A. (Eds.). Road User Behavior: Theory and Practice.
Van Gorcum, Assen.
Regan, M.A. and Horberry, T. (2004). A driving simulator evaluation of enhanced road
markings. In Proceedings ofAustralasian Road Safety Research, Policing and Education
Conference, Australia.
 21. Present and Future of Simulation Traffic Models 427

Rizzo, M., Jermeland,1. and Severson, 1. (2002). Instrumented vehiclesand driving simu-
lators. Gerontechnology, 1(4),291-296.
Salvucci,D.D.,Boer,E.R. andLiu,A. (2001).Toward an integratedmodelof driverbehavior
in a cognitivearchitecture. Tranportation Research Record No. 1779,pp. 9-16.
Schvaneveldt, R.W., Reid, G.B., Gomez,R.L. and Rice, S. (1998).Modellingmental work-
load. Cognitive Technology, 3, 19-31.
Spillers, F. (2003). Task analysis through cognitivearchaeology. In D. Diaper, N. Stanton
(Eds.). The Handbook ofTask Analysis for HCI. LaurenceErlbaumAssociates, Mahwah,
N1.
Stone, P. and Veloso, M. (2000). Multiagent systems: A survey from machine learning
perspective. Autonomous Robots, 8(3), 345-383.
Sukthankar, R., Hancock,1.,Pomerleau, D. and Thorpe,C. (1996).A simulationand design
system for tactical drivingalgorithms. In Proceedings ofAI, Simulation and Planning in
High Autonomy Systems.
Sukthankar, R., Hancock,1. and Thorpe,C. (1998).Tactical level simulationfor intelligent
transportation systems.Journal on Mathematical and Computer Modelling, Special Issue
on ITS, 27(9/11), 228-242.
Sukthankar, R., Pomerleau, D. and Thorpe, C. (1995). SHIVA: Simulated highways for
intelligent vehicle algorithms. In Proceedings ofIEEE Intelligent Vehicles.
Theeuwes,1. (2002). Sampling information form the road environment. In R. Fuller, 1.A.
Santos (Eds.). Human Factors for Highway Engineers. Elsevier, Amsterdam.
Theeuwes,1. and Hagenzieker, M.P. (1993).Visualsearchof traffic scenes:On the effectof
location expectations. In A. Gale, LD. Brown, C.M. Haslegrave and S.P. Taylor (Eds.).
Vision in Vehicles (Vol. 4, pp. 149-158). Amsterdam, The Netherlands.
Toftin, D., Reymond, G., Kemeny, A. and Droulez, 1. (2003). Influence of steering wheel
torquefeedbackin a dynamicdrivingsimulator. In Proceedings ofthe Driving Simulation
Conference, North America,Deaborn.
Zhang,Y., Owechko,Y. andZhang,J. (2004).Drivercognitiveworkloadestimation: A data-
drivenperspective. In Proceedings of 77th International IEEE Conference on Intelligent
Transportation Systems, Washington, DC.
Zhang, Y., Owechko, Y. and Zhang, J. (2004). Learning-based driver workloadestimation.
In Proceedings of 7th International Symposium on Advanced Vehicle Control, Amhem,
The Netherlands.
Zijlstra,F.R.H. (1993).Efficiency in Work Behaviour: A Design Approachfor Modern Tools.
Delft University Press, Delft.
Index

A Advanced driver assistance systems

(ADAS)/Driver assistance systems (DAS),
ABS. See Antilock braking systems 8,35
ACC mental model, 131 Adverse behavioural consequences, 149
ACC. See Adaptive cruise control Affect intensity measure, 183
Accident cause Affect-conditioning history, 181
ABS as a potential reducer, 216-18 Affective response, 181-82
importance of driver, 3 Affordances of car/driver system, 197
Accident rate, 19-20 AGILE project (QLRT-2001-00118),
Accidents, types of, 217 15
Account of feelings, 225 AIDE (Adaptive integrated driver-vehicle
ACME driver model, 34-35 interface) ED-funded project
ACT-R driver model, 30-32 adaptive forward collision warning systems,
Action boundary, estimate of, 193 112
Action goal structure for changing gear, behavioural adaptation, 157-58
109t behavioural effects of driver support
Activated knowledge, 255 functions, 62
Active driving, 192 identifying crucial behavioural adaptation
Adaptive automation, 50 issues, 149
Adaptive behaviour, 77, 80, 348-49 AIDE (IST-I-507674), 22
Adaptive control models Aijzen, I, 209-10
behavioral changes, 87 Algorithmic description of situation awareness,
driver-vehicle-environment, 91-93 254-58
operation of driving task, 107 Anticipation agent, 326
Adaptive cruise control Anticipation, 318
behavioural impact of, 150-51 Anticipatory open-loop control, 286-87
driver behaviour issues, lOt Anticipatory schemata, 93
evolution of, 412 Antilock braking systems
interactions between a driver and automation, definition of, 19
design of, 52-54 effects of, 214-18
investigating learning effects, 130-31 Approximate time frames, 89
as a partial automation, 52 Artificial intelligence (AI) principles, 3
as a support while driving, 71 Attention capture by competent stimuli, 183
system methods, 391 Attitude(s)/personality (ATT), 114,355,357,
Adaptive function allocation, 50 365,419
Adaptive, 355,421 Attitude toward the behaviour, 168
ADAS design and impact assessment, 20-22 Automatic vehicle guidance (AVG) system,
ADAS-improve driving safety, 9-14 12

429
430 Index

Automation surprises, 50, 52, 55 Car driving

Automation, 10, 11 cognitive view of, 85-86
Automotive technology, technological advances, definition of, 3
147 emotional experience, 214
Autonomous (intelligent) agents, 405, 410, Car-following and risk, 143
422 Car-following situation-risk
Available friction in a curve, 193 safety margins, 150-51
Available time, 193-96. See also Driver short-time headways, 154
workload speed error, 293
Average speed and/or speed variability-relation Categorization, 328
to accident risk, 142 Changes in drivers' activity, 147-48
Avoidance behaviour, 13 Choice of speed/Speed choice. See also Driving
Avoidance learning, 166 within safety margins
AVVA~(IST-2000-28062),20-22 driver's behavioural approach, 165-67
GSR activity, 190
ChromograninA (CgA) in saliva, role in driver's
B
mental tension, 46-47
Badly designed in-vehicle (or roadside) Closed system, 90
supports, 136 Cockpit activity assessment (CAA), 9
Behaviour analysis, 165-67 COCOMIECOM framework, 61, 77,80-81
Behaviour and risk, link between, 142-43 Cognition in control, 93
Behavioural adaptation Cognitive abilities, 86
advanced driver support systems, 149-57 Cognitive approach, 86
in AIDE project, 157-58 Cognitive architecture, 31
definition of, 61, 75 Cognitive complexity, 124
mismatch between technology and human Cognitive distraction, 259-60
capability, 219-20 Cognitive operations, 181,213
and risk compensation, 210-11 Cognitive resources of driver, 125
Behavioural adaptation formula, 14 Cognitive systems engineering (CSE), 91
Behavioural adaptation to driving support Collision, 171-72
functions Collision-avoidance system, impact of, 143
countermeasures against, 144 Comfort zone, 199
definition of, 148 Common modular framework, 40, 41f
proposed demonstrations, 75-76 Comparator, 212-13
Behavioural approach to driving, 87 Compensatory closed-loop control model, 287
Behavioural effects and actual accident risk, Compensatory steering, 92
relation between, 78-80 Complex task, 350, 352t, 411t
Behavioural effects and road safety, Comprehensive driving model, ideal
conceptualising relations between, 63 characteristics, 39-40
Behavioural effects of driver support functions, Conditional probability distribution, 46
74-78 Consistent goal state, 67
Behavioural events while approaching a STOP Constraint-satisfaction process, 257
sign, 45 Construct-action-event cycle, 69
Behavioural impact of ISA, 150 Contention scheduling, 257
Behaviour-based human environment creation Context -dependent choice reaction task, 262-63
technology project, 44-47 Contextual control model (COCOM), 68-70.
Beliefs about behaviours, 168 See also Cyclical model of control
Brake reaction time, 194 Control cycle, 93-96
Control goals, 99
Control measures in on-road driving, 198
c Control model, 369-73
Car accidents to human error, 43 Control tasks, 71
Car control, 192 Control theory/cybernetics, 68, 89-91, 277
 Index 431

Control tracking, 99 Driver as an active agent, 77

Control, 67 Driver availability estimator (DAE), 9
Control-level behaviour, 113 Driver behaviour as goal-directed activity,
Control-level information, 112 67
Controlling tasks, 277 Driver behaviour issues when introducing
Cooperative (systems), 341 advanced cruise control, lOt
Costlbenefit model, 364 Driver behaviour models
Critical avoidance manoeuvres, 218 classification of, 87
Crooks, L.E., 95 in ED and international projects, 15-22
Crossover-model, 281 existing models, 63-66
Curve braking, 100 motivational aspects in, 209-10
Cyclical model of control, 93. See also Driver behaviour
Contextual control model (COCOM) cognitive aspects as determinants, 210
cognitive control loop, 98-99
concurrent goals, 96
D
definition of, 67
Damasio, A.R., 179 drivers with ABS cars, 218
Dangerous driving, 183 motivational models of, 5-6
Data sets, 28, 32, 390 Driver behavioural changes
Deceleration and acceleration schema, link diversity of, 150-52
between, 257-58 effects of ITS, 219, 221
Decision, 8f, 13-14, 34, 47, 86, 346 short-and long-term changes, 10
Decision authority temporal dimension of, 156
adaptive automation, 50-52 Driver capability, 172
between driver and automation, 44 Driver characteristics (DC) module, 9
Decision/Decision making systems, 184, 211, Driver competence, 167
249,351,423 Driver component, dimensions of, 37
Degree of penetration or use rate of device, Driver control behaviour, 93
135 Driver control of vehicle, 112
Delayed avoidance, 182-83 Driver decision making
Desensitization on approach to an intersection, 108t
to emotional feedback, 177 closed-loop homeostatic system, 169-70
to potential threats, 167 feelings of risk, 183-84
Design of in-car HMI, 125 misinterpretation of, 112-13
Design of interactions between a driver and Driver distraction
automation, 52-54 definition of, 76-77
Designing a model, basic requirement, 112 types of, 258-59
Desired lane position, 369 Driver impairement level (DIL), 365,419
Desired speed profile, 368f, 369 Driver inattention, 76
Destination entry while driving, 129-30 Driver in control model, 108t
Detecting speed changes, 227-28 Driver model
Detection rate, 262, 263 architecture, 27
Developing an online model, 117-19 basic architecture of a simulation, 36-38
Dialog errors, 128 model dimensions, 37, 38f
Difficulty of driving task, 173-74 model toolbox, 38-39
Direct (side) behavioural effects, 75 as technical controller, 288-89
Discrete-event model, 43-44 verifying the model, 116-17
Distance from crash, 192 'Driver simulator' software, 39
Distributed artificial intelligence, 405 Driver state
Diversity and variability of road situations, descriptive driver behavior, 355-57
152-53 monitoring of, 105
DRIVABILITY index (DI), 9, 21 project concerning, 48
DRIVABILITY Model, 6-9 role in error generation, 365
432 Index

Driver state (cont.) Driving performance and risk (for an individual

significance in driver model, 115-16 driver), relation between, 78-79
significance in driver response to an alert, 53 Driving performance indicators, choice of, 150
Driver state degradation (DSD), 9 Driving performance, 76
Driving style, 153-55 Driving support functions
Driver support functions-characterization with functional characterization of, 62
respect to the goal(s), 72-75 role of, 67
Driver support systems, 85 Driving task
Driver training and assessment, driver models demands, 172
for, 15 driver performance of, 113-15
Driver workload. See also Available time hierarchical structure of, 88
estimating task demand, 118 immediate goals, 178
reduction by automation, 9-11, 20 potentially useful skills, loss of, 136
Driver, 3 Driving task levels, 66
Driver's attitude, 153 Driving task modeling, 22-23
Driver's choice of speed and safety margins, Driving task-descriptive models, 106-9
96 Driving under more difficult conditions, 141
Driver's choices between options, 197-98 Driving with supports-behavioural factors,
Driver's complacency, 54 136-37
Driver's fatigue, estimation of, 47 Driving within safety margins, 95. See also
Driver's mental model, 93-94 Choice of speed/Speed choice.
Driver's mental tension, estimation of, 46-47 Driving-related inattention to the forward
Driver's model, 407-8, 412-14 roadway', 76
Driver's normative (or baseline) behaviour, 45 Drunk driving
Driver's overtrust in 'smart and reliable' accident risk, 79
automation, 44 risk monitoring function, 226-27
Driver's situation recognition, 48-50 Dual control problem, 94
Driver's state, estimation of, 46-47 Dual mode model, 287
Drivers' acceptance of support system, 150, Dual-carriageway situation, 182
152-53 Dynamic interactions, 66, 348--49
Drivers' interaction situations, 152-53 Dynamic route guidance (DRG) systems, 20
Drivers' interactions with their driving Dynamical representation of driver behaviour,
environment, 147 67-68
Drivers' perception of motorcyclists and
pedestrians, 228
E
Drivers' psychological states and their
transitions, 53 EC projects dealing with driver training, 15
Driver-specific factors, 4 Effects of ITS-monitor model for prediction,
Driver-vehicle-environment (DVE), 91, 347, 227-29
384,417-23 Efficiency-thoroughness trade-off (ETTO)
Driving strategy, 70
behaviour, 95-99 Electro-hydraulic system, 101
cognitive view of, 85-86 Electronic stability control (ESC), 220-21, 227
destination entry during, 129-30 Elementary functions, 350-52, 364
hedonistic motives and pleasures, 201-2 Elementary task, 350-51, 352t, 411
as a self-paced task, 175 Elementary utility models, 138
upper limits of tolerance, 179 Emergency-braking alert, driver's response to,
Driving as information processing, 111t 53-54
Driving behaviour modelling-Bayesian network Emergent behaviour, 423
model, 45 Emotional reactivity, individual differences in,
Driving behaviour, 45,239,330,413-14,422 182
Driving goals and context, 6 Emotional responses arising from the unfolding
Driving moods, 201-2f road and traffic scenario, 183
 Index 433

Emotional signals, 179-80 Good drivers, 94

Emotionally competent stimuli, 179 Good progress of trip, 201
Emotions, 179 Guidance level of driver model, 288
Enforcement of traffic regulations on driver
behaviour, psychological effects of, 167
H
Engineering control applications, 68
Engineering estimate lor intrinsic effectiveness Hankey, Jon, 29-30
estimate, 135 HASTE ED-funded project, 77-78, 118
Environment affords, 196-97 HASTE project, 118
Episodic memory representation, 256 Helmet-wearing laws, 17
Erroneous safety margin setting, 79 Hierarchical control models, 87-89
Error analysis, 128 Hierarchical layers of control, 98-99
Error reduction in all error categories, 128 Hierarchical models, 66
ESC and ABS utilises technologies, 221 High anger drivers, 183
European Commission (EC), 379 High-level motivation (on the targeting level), 75
Expected safety effect, of a device, 135 HMI design (AWAKE, COMDNICAR, AIDE,
Experience/Competence (EXP), 355-57, 365 etc.),23
'Expert users' as subjects, 132 Homeostasis
Extended control model (ECOM), 66, 70-72 basic physiological needs, 209
Extraversion and sensation seeking, personality definition of, 96
traits of, 183 Human (or operator) modeling, 3
Human (risky) choices-role of emotions, 202-3
Human behaviour, 3
F
Human controller/Human abilities, 68, 86
Familiarization stage, 156 Human controlling behaviour, 277-83
Feed forward (or anticipatory) control, 68 Human errors, 43, 124,341,349
Feedback (or compensatory) control, 68 Human factor (s)-information processing, 64,
Feedback to driver, 197 110-11f
Feelings, 223 Human factor variables, 172-74
Field of safe travel, 95 Human information-processing (HIP) models,
Frequency or cost of insurance claims, 20 90
Fuller, Ray, 13,96 Human performance, 66
Functional balance, 222 Human-machine interaction, 344, 356t, 380,
Functional models, 87 383-5
Functions supporting non-driving tasks, 61 Human-machine interface (HMI), 123
Fuzzy correlations, 358-62, 366 Hypovigilance diagnosis module (HDM), 381
Fuzzy logic based motorway simulation
(FLOWSIM), 35
Fuzzy rules, 358, 362
Identifying candidate parameters, 117
Improved braking (i.e. longitudinal tracking)
G
performance, 75
GADGET-matrix, 5-6, 7t Inadequate adaptation, 80
Generating driver models, 27 'Inattention blindness' or
Generation of extra mobility (VMT), 141-42 'looked-but-did-not-see' , 259
Genetic algorithms, 3 Increased workload, 229-30
Gibson, J.J., 209-10 Increasing drivers' attention, 227-29
Goal, 81, 85, 209, 254, 422-4 In-depth accident studies, 96
Goal-directed activity, 67 (In fact re-training), 15
Goal-directed behaviour, 67 Indirect (side) behavioural effects, 75
Goal-mean task analysis, 411 Individual differences in affective response, 182
Goal-oriented action, 281 Individual resources, 7
Goals for life and skills for living, 6 Individual support systems, impact of, 150
434 Index

Informal rules or behavioural norms-drivers' use In-vehicle information system (lVIS)/Driver

and acceptance of assistance, 152 information systems (DIS), 29-30, 379
Information processing In-vehicle information system (lVIS)/Driver
driving task definition in terms of, 110 information systems (DIS) and situation
element of driver model architecture, 27 awareness, 260
models, 64, 111f Inverse square law of affect intensity, 182
modes of, 225-26 ISA. See intelligent speed adaptation
significance in driver modeling, 317-19 ITS. See Intelligent transport system
steps of, 237 IVIS tasks, 260-62
sub-model for ACME driver model, 34 IVIS, principle of, 30
systems, 254
Information processing and decision making,
routes or modes of, 225-26
J
Information processing models, 64-66 Jerky curve with 'flats', 278
Information processing system, 254, 345 Joint cognitive system (JCS), 68-69
Information processor, 34 Joint driver-vehicle system (JDVS), 69
Infrastructure component, dimensions of, 37 Joystick controlled cars, 100-1
Instrumental learning, 166-67 Joystick drivers, 101
Integrated Projects, 23
Integration phase, 158
K
Intelligent speed adaptation
driver's interaction with other road-users, 153 Kickback in ABS, 219
performance of driving task, 149 Knowledge bases, 326-27
Intelligent transport system-risk monitoring Knowledge/skills level, 7
ability, 227-29 Knowledge-based behaviour, 4
Intention(s) Knowledge-based performance, 66, 266
detection of, 396-98 Knowledge-based, 237, 266
algorithm, 393-95 Kondo, M., 284
formulation, 355
model based on definition of, 351
L
recognition, 47-54
significance in driver model, 115f, 116f, Lack of tactile feedback, 100
363-66 Lag equalization, 309
significance in lane change algorithms, 413 Lane/road departure risk, 392
understanding and communication, 55 Lane-keeping performance and risk, 142
Intentional stance, 67 Lapses, 266-68
Inter-individual differences, 132 Lateral acceleration, 193
Internal mental processes as causes of overt Lateral control driver models, 284-86
behaviour, 167-68 Lateral control, 238-40, 284
Internal states and external behaviour, Lateral placement, 239-44
relationship between, 222 'Law of requisite variety', 93
Interpretation Learnability of a system, 124
of a driver, 48,51, 112,256 Learnability of driver assistance, 125-26
as an element of PIPE driver model, 22, Learner-oriented design process, 124
408 Learning, 124
estimation, 393-95 Learning and appropriation phase, 158
risk, 78 Learning experiment-examples, 126-32
of signs and symbols, 353-55 Learning processes
Intersection Collision Avoidance Warning to drive with new driver support systems,
System, 235 155-57
Intimate zone, 191-2 and leamability in the in-car domain, 131-32
Introducing ACC, driver behaviour issues, lOt Less attention, 137
In-vehicle functions, 61-62, 73, 74 Level II situation awareness, 255
 Index 435

Level of automation (LOA), 51-52, 54 Mistakes, 266-68, 271

Level of description-TCI model, 174 Model calibration and validation, 28-29
Level of emotional tension or anxiety, 177, 190 Model components, hierarchy of, 36f
L-H Interactions, 418 Model of basic indicators of driver operational
Life-saving equipment, purchase of, 180 navigation model (BIDON), 365
Limited capacity metaphor, 77 Model quality, 332
Limits to closeness to hazard, 196, 198 Modelling human behavior, 344-45
Linear driver models, 283-89 Modelling driver behaviour, 85
Liveware (L): L-L interactions, 418 Modelling steering behaviour, 91-93
Load centre position (LCP), 47, 48 Models of performance, 316
Local final state, 321 Monitor Model, 222, 224-27
Local initial state, 321 Monitoring, 71
Locus of control, 11-12 Motivational models
Longitudinal control, 240-50 of driver behaviour, 64-65
Long-term goal, 70 of driving task, 109-12
Looked-but-did-not-see, 259 Motives, definition of, 208-9
Looming stimulus, responses to, 182 Motor abilities, 86
Loss of control of driving task, 171 Motorcycle helmet, use of, 17-19
Multi-agent approach, 421
Multi-agent distributed systems, 401, 420
M
Multi-agent systems, 420
Machine intelligence given decision authority, Multimodal HMI, evaluation of, 127-29
51f Multiple comfort zone model, 198-201
Macroperformance, 315 Multiple resource theory, 64
Main traffic risk condition detection, 391-93 Multiple task sharing, 64
Manchester driver behaviour questionnaire, 268 Multitasking while driving, effects of, 76-78
Manipulation of speed, 175
Manoeuvring level of driving, 112
N
Man -plus-the-controlled-element, 345
Manual control models, 63-64 Naatanen, R., 109-10
Manual control, 91, 293-97 Navigation, 63, 66, 92, 125, 236-37
Mastery of traffic situations, 6 Near-accident(s), 235
Matlab, 39 Negative side effects of behavioural adaptation,
Mean-free driving speed, 245-46 150
Means to accelerate learning process, 156 New Energy and Industrial Technology
MECTRON-HMI group at university of modena Development Organization (NEDO), 44,
and reggio emilia driving 45
simulator, 414-17 Nomadic devices, 61
Medium-term scale goal, 70 Non-driving-related functions, 74
Mental model(s), 320-22, 317, 332, 338-41 Non-normative behaviour, detection of, 45--46
Mental processes needed for driving, 34 Non-stationary behaviour, 282-83
Mental simulation, 324, 330, 337 Normal (controlled) task performance, 71
Mental workload, 77 Normal Driving, 367-69
Message sending, 325 Normal driving-compensatory and anticipatory
Michon's hierarchical control model, 4-5 control, 97
Microperformance, 315 Normative approach to driving, 87
Microscopic traffic simulator at the University of Normative behaviour, simulation approach for,
Michigan, 413 349-52
Ministry of economy, trade and industry Normative driver Behaviour, 350, 353-55
(METI), government of Japan, 44, 45 Novice and experienced drivers
Ministry of education, culture, sports, science interpretations of current situation,
and technology (MEXT), government of 256
Japan,44 visual scanning between, 182
436 Index

o Position controller at stabilization, 288-89

Potential differential impact of support systems,
Occlusion, 238
153-55
OECD (Organisation for Economic Cooperation
Potential hazard scenarios, 184, 194
and Development) report
Predictive analysis, 374
On-road research, need for, 203
Predictive DVE interactions, sample cases,
Open loop-closed system versus closed
371-74
loop-open system, 90f
Predictive model (s), 112, 316
Open system, 90
Predictive tool, 116, 347
Operation of driver information and driver
Preferred workload, 178
assistance systems, 125
Prevention of excessive speeds, 227
Operational features of vehicle being driven, 172
Primacy, 236
Operational module, 318
Primary control of car, 98
Optic flow field, 244
Primary emotions, 223
Optimal control model, 32-34
Principle of preservation of the accident rate, 12
Orienting reflex, 226
Proactive safety technology, 43
'Out-of-the-loop familiarity,' 11
Probability distribution, 45-46
Overall driving task and identifying behavioural
Procedural knowledge, 172, 320
changes, 148
Production models, 107-9
Overtaking case
Proposed structure, 115-16
drivers' monitoring ability, 228
Proxemics approach, 191-92
identifying driver intention, 118-19, 156
Proximity bubbles, 323
Overtrust on a system, 11
Psycho-motor dimensions, 38, 39f
Pursuit control, 92
p

Parallel user group, 39 Q

Parameters, 356-58, 365-67, 423-24 Quasi-linear model, 280
(Partially) automated supports, 137
Path errors, 238
R
Perceived behavioural control, 168
Perceived road curvature, 92 'Radar web' graph, 37
Perception Rage, 183
based on visual information, 423 Railway grade crossings, 249-50
as a behavioral entity, 90 Rapid deceleration rate, 54
of a driver, 407-9, 414 Rasmussen, J., 237
of knowledge and skills, 18t Rasmussen's division of operative behaviour, 5
signal, 22 Real-time model of the driver, 112-17
risk,8,202 Recognition errors, 128
Perception module, 326 Reducing driver workload, 9-10
Perception, interpretation, planning and Reducing speed, 97
execution (PIPE), 22, 407-9 Reduction of recognition errors, 128
Perceptive exploration zones, 321 Regulating, 71
Perceptive exploration, 319-21, 327-32 Regulation-compliant behaviour, 167
Perceptive queries, 319 Reinforcement theory, 181
Perceptual abilities, 86 Relationships between driver arousal (GSR
Perceptual cycle, 93 level), 177
Perlormance, 266-67, 309-10, 315,408t Relative zones, 323
Peripheral detection task (PDT), 261 Remarkable points, 323
Permanent tasks, 351-55 Remnant, 278-79
Physical environmental factors, 172 Rewards and punishments, 13, 109-10, 181
Physical workload- muscular output, 177-78 Risk (as a function of performance), 79
Planning, 157,407-8, 410t Risk and risk perception, 137
PLATO spectacles, 238 Risk assignment criteria, 391-2
Population-based incremental learning, 424 Risk avoidance model, 111t
 Index 437

Risk awareness, 8 Safety

Risk compensation, 13, 210 control schemes, 54
Risk contributor, 389 interventions, 170
Risk feeling, 176f margin management-threshold model,
Risk finding and avoidance project, situation and 199-200
intention recognition for, 47-54 margin threshold, 75
Risk homeostasis theory Safety margins
definition of, 12 car-following situations, 150-51
driver's choice of speed, 169-71 in on-road driving, 198
target risk level, 211-4 role in driver behaviour, 65
Risk measures in traffic safety studies, 139-40 for survival and control, 199-200
Risk monitor, 190, 198 Safety zone, 191-93
Risk taking, 79 SALSA, 260
Risk threshold model, 110t, 315 Salvucci. Dario, 30
Risk Satisficing, comfort through, 201-3
factors, 18t Scenario-assessment unit, 382
models, 315-16 Schema, 257
perception and awareness of, 5-10 Schemata, 257
potential, 331 Scientist/researcher group, 39
Riskier behaviours by drivers, 137-39 Seat belt accident statistics, 140-41
Risk-level assessment unit (RLA), 383-84 Seat belt wearing rate, 140-41
Risk-taking behaviour, 15, 135-45 Seat belts, 15-17, 137-38
Risky behaviour, 212, 344, 374 Secondary emotions, 223
Road safety research-behavioural adaptations, Secondary task demand, 118
148-49 Secondary tasks while driving, 76
Road traffic accidents, 268-71 Security 190,402,421
Road traffic simulation survey, 401-3 Self-explanatory' support system, 125
Road traffic, 210 'Sensation Seeking' and/or 'locus of control'
Road transport, 165 (LOC),154
Road use by less qualified segments of the Sensitivity of experienced drivers to prevailing
driving population, 141 traffic conditions, 182
Road users' behavioral dynamics, 203 Sensitivity to mental workload, 178
Rothengatter, T., 209-10 Servo-control model of drivers' steering
Routine driving, unifying emotional concepts in, behaviour, 92-93
198-201 Servo-control models, 91-93
Routine open road driving, 194 Shaft model, 284
Rule following, 200 Short-term goals, 70
Rule-based approach, 389 Signalised intersection, 200, 249
Rule-based behaviour, 4 Simple simulation of driver performance
Rule-based performance, 266 (SS-DRIVE),348-70
Rule-based system for TERA algorithms, Simple task condition, 333
388-90 Simulation, 400-5, 407, 412-14, 417-23
Rule-based, 237, 266 Simulator environment, 406
Rules, 266-68, 362, 384, 408t, 418 Simulators, objective of, 422
Sinusoidal reference input signals, 279
s Situational awareness, 136, 410, 414
Situation awareness
Safe driving algorithmic description of, 254-58
cognitive and psychomotor skills, 99 as an adaptive behavior parameter, 355, 356t
preconditions of, 253 association with in-vehicle information
role of driver support systems, 85 system tasks, 260-64
Safe driving behaviour, goals for, 95-96 comprehension based model of, 258-60
'Safety critical' scenarios, 150 definition of, 253
Safety effects-new technology in vehicles or in driving models, 253-58
road infrastructure, 135-36 effect of IVIS tasks, 260-62
438 Index

Situation awareness (cont.) Storage structures, 317

methodology for study, 332-37 Strategic module, 318
of a driver, 43,48, 114-18, 320, 323 Studded tyres, 19
relevance in HASTE project, 118 Subject testing, 123
variables and fuzzy values, 361-63 Subject's learning state, 123
Situation awareness global assessment technique Subjective norm, 168
(SAGAT), 260 Subjective risk levels, 4-7t
Situation model, construction-integration Subjective risk monitor, 203
theory, 255-57 Subjective workload, 151, 155
Situational context-role in behavioural changes, Subjectively chosen safety margins, 65
152-53 Summala, Heikki, 95, 110
Situational performance, 315 Supervisory attentional system (SAS), 258-60
Situation-appropriate actions, selection of, Support for monitoring, 73
257-58 Support for regulating, 73
Situation-specific representation, 256 Support for targeting, 73-74
Skill-, rule-and knowledge-based behaviour Support for tracking, 72-73
(SRK) model, 88 Support system's competence, 155
Skill-based behaviour, 4 Supra-threshold feelings of risk, 184
Skill-based level of performance, 266-67 Sustainable workload for a cognitive task, 178
Skill-based, 126, 237 Sustainable workload for a physical task, level
'Skill models' of driver behaviour and safety, of, 178
189 System
Sleep apnoea and fatigue-accident risk, 228 as an abstract construct, 90
Slipperiness, 193 learnable, 125
Slips, 266-67 System boundaries, 91
Social consequences of behaviour, 168 System safety, 184
Somatic marker strength, 181
Somatic-Marker hypothesis, 179-83,221-24.
T
See also Threat-avoidance theory/model
Spare capacity, lOt, 173-74 Tactical driving frames, 340
Spatiotemporallimits of the driver-car unit amid Tactical mental representation, 329, 336
of traffic, 197 Tactical module blackboards, 325-26
Speed control model with accelerator pedal, Tactical module, 318-32
299-302 Take-over situations
Speed control, 16t, 304-6, 370 drivers intervention and behaviour, 156
Speed limit, 172-73 evaluation of ADAS, 130-31
Speeding behaviours, 169, 220 Target feeling, 214-5
Speeding, 153, 183 Target risk, 169-70, 214
Standard operating procedures (SOPs), 184 Targeting control level, 71
State of control, 178-79 Task analysis models, 87
State transitions, 53 Task demand, 114-18, 171-76, 183, 356,
Statistical risk ratings, 176-77 358-65
Steering angle, 285, 287-88 Task difficulty level, 175-78
Steering behaviour, 118 Task difficulty
Steering control, 38, 92-93, 100, 238, 296-97, determinants of, 173-4t
303-4 drivers' motives, 190-91
Steering system, 92 elements of, 172
Stimulus and response, effect relation between, Task difficulty homeostasis, 96, 176-79
91 Task environment, 254
Stimulus conditions (discriminative stimuli), 166 Task models, 106
Stimulus generalization, 166 Task, 350-52
'STI' -model, 285 Task-capability interface (TCI) model, 171-79
Stopping deceleration, thresholds for, 200 Tau(r),65, 193,286
 Index 439

Taxonomic models, 87 Traffic simulator models, overview of, 400-1

Temporal factors affecting behavioural Traffic system
adaptation, 158 composition of, 3
Tendency to seek enhanced external stimulation, time and space, role of, 196-98
183 Traffic, 406,410-14,419-24
TERA. See Traffic and environment risk TRAINER project (GRDI-1999-10024), 15
assessment Transportation Research Board, Washington,
Testing stage, 156 DC, 84th annual meeting of, 26
Theeuwes, 1., 237 TTC. See Time-to-collision
Theory of planned behaviour (TPB), 167-69 TTCbr, 240, 243f, 244-5, 248
Threat-avoidance theory/model. See also TTC gas, 247--48
Somatic-Marker hypothesis TTCmin, 240--48
drivers' speed 'choices', 166 TTCstim, 247--48
subjective probability (or likelihood) of an Tustin, A., 277
accident, 13 Tustin-model: Linear part + remnant, 277-79
Threshold model for safety measures, 191
Time considerations-cognitive control cycle,
97-98
u
Time constraints, 89 Unbelted drivers, 15
Time in traffic behaviour, role of, 194 Uncertainty, level of, 97-98
Time lag problem, 100-1 Unconscious in psychoanalysis and
Time line analysis, 350 psychodynamic theory, role of, 209
Time margins, 192 Unintentional skidding, 94
Time pressure, 97 Unlearned reflex responsesl'learning readiness',
Time-to-collision, 96, 192-3, 235, 238--48, 182
391 Upper limit of competence of driver, 172
Time-to-collision or time-to-contact, 192-93 Urgency, 194,237
Time-to-intersection (TTl), 248--49 USA accident statistics, 17
Time-to-line crossing, 65, 192, 235, 238-39 User's profile configuration agent, 422
Time-to-object information, 65 Utility maximisation model, 14
Time-to-stop-line (TTS), 249-50
TLC. See Time-to-line-crossing
Total task times of destination entry dialogues,
v
129 Vaa, TroIs, 210
TOTE (test-operate-test-exit) mechanism, 179 van der Horst, Richard, 32
Tracking control, 71 ve-DYNA advanced driver controller,
Tracking-models, 283 288
Traffic, 406,410-14, 419-24 Vehicle behaviour, 283
Traffic and environment risk assessment (TERA) Vehicle component, dimensions of, 37
architecture, 386-88 Vehicle configuration, 409
implementation, 395-97 Vehicle manoeuvring, 6
role of, 9, 385 Vehicle subsystems, 407-11
Traffic conflict(s), 235 Vehicle/road system, 200
Traffic demands, 97 Vehicle's trajectory and speed, 172
Traffic flow on a main road, 196 Vehicle-actuated control, 249-50
Traffic hazard, 176 Vehicles with a driving recorder system,
Traffic psychology-behavioural adaptation, 189 45
Traffic risk estimation module (TRE), 381-83, Vestibular stimulation, 182
398 Violation,7t, 18t, 79, 267-72
Traffic simulator Violations of safety margins, 79
applications, 412-16 Visibility, 246--47, 358-59, 396f, 41H
case studies, 405-7 Visual and working memory demands of IVIS
types of, 403-5 tasks, 262-63
440 Index

Visual fixations, patterns of, 181-82 Workload management meta-functions, 74

Visual or motoric demand, 124 Workload reduction
Vocabulary errors, 128 driving with advanced driver support systems,
150
role of adaptive cruise control, 52
w
Workshop content/themes, 27
Warning level index, definition of, 383t
Warning strategies unit, 383
Whispers of affect, 184
z
Wilde, G.C.S., 96 Zero risk, 214
Working memory resources, 259 Zero-risk hypothesis, 13, 64
Workload (mental), 416 Zero-risk model, 166
Workload management functions, 61, 74 Zuckerman, M., 183