Modification of Vehicle Handling Characteristics via Steer-by-Wire

Paul Yih Jihan Ryu J. Christian Gerdes

Design Division Design Division Design Division

Dept. of Mechanical Engineering Dept. of Mechanical Engineering Dept. of Mechanical Engineering
Stanford University Stanford University Stanford University
Stanford, CA 94305-4021 Stanford, CA 94305-4021 Stanford, CA 94305-4021
Email: [email protected] Email: [email protected] Email: [email protected]

Abstract demonstrate that active steering control can achieve

greater driving stability than differential brake control.
This paper presents a physically intuitive method for
altering a vehicle’s handling characteristics through Although feedback of sideslip angle for active steering
active steering intervention. A full state feedback control has been proposed theoretically [5], the
controller augments the driver’s steering command via difficulty in estimating vehicle sideslip presents an
steer-by-wire to achieve desired handling behavior. obstacle to accomplishing this in practice. Stability
Accurate estimates of vehicle states are available from control systems currently available on production cars
a combination of Global Positioning System (GPS) typically derive slip angle from sensor integration or a
and Inertial Navigation System (INS) sensor physical vehicle model, but these estimation methods
measurements. By canceling the effects of steering are prone to uncertainty [6]. Because sideslip is
system dynamics and tire disturbance forces, the steer- extremely important to the driver’s perception of
by-wire system is able to track commanded steer angle handling behavior, quality of the driving experience
with minimal error. Experimental results verify that depends strongly on quality of the feedback signal.
with precise steering control and accurate state While this dependence is less critical for stability
information, a vehicle’s handling characteristics can control systems—which tend to engage when the
be modified to match driver preference or to vehicle is already undergoing extreme maneuvers—to
compensate for changes in operating conditions. improve handling behavior during normal driving
requires cleaner and more accurate feedback.

1 Introduction A new sideslip estimation scheme combining GPS and

INS sensor measurements overcomes many of the
As a step toward fully integrated vehicle dynamic drawbacks of previous estimation methods [7]. For
control systems, active steering capability will be this paper, a test vehicle converted to steer-by-wire is
available on select production vehicles within one or used to demonstrate that a vehicle’s handling
two years. The potential benefits of active steering characteristics may be find-tuned through a
intervention, particularly to improve handling combination of GPS/INS feedback and precisely
behavior during normal driving, have received controlled active steering. The first part of the paper
considerable attention from both the automotive briefly discusses the estimation scheme along with a
industry and research institutions. As early as 1969, physically motivated approach for full state feedback
Kasselmann and Keranen [1] developed an active control of an actively steered vehicle. The latter part
steering system based on feedback from a yaw rate of the paper describes the design of the steer-by-wire
sensor. More recent work by Ackermann [2] system that provides active steering capability to the
combines active steering with yaw rate feedback to test vehicle. Experimental results clearly show the
robustly decouple yaw and lateral motions. This change in handling behavior achieved with full state
method is effective in, for example, canceling out yaw feedback steering control. In addition to matching
generated when braking on a split friction surface. In handling behavior to driver preference, the system
[3], Huh and Kim devise an active steering controller successfully counteracts handling differences caused
that eliminates the difference in steering response by shifts in weight distribution.
between driving on slippery roads and dry roads. The
controller is based on feedback of lateral tire force
estimates derived from vehicle roll motion. Most 2 Planar Bicycle Model
recently, Segawa et al. [4] apply lateral acceleration
and yaw rate feedback to a steer-by-wire vehicle and A vehicle’s handling dynamics in the horizontal plane
are represented here by the single track, or bicycle
model with states of sideslip angle, β, at the center of to implementing an active handling system with full
gravity (CG) and yaw rate, r. state feedback control. Although yaw rate data is
available on many production cars from rate
gyroscopes, sideslip cannot be directly measured and
must be estimated instead. Two common techniques
for estimating this value are to integrate inertial
sensors directly and to use a physical vehicle model.
Some methods use a combination or switch between
these two methods appropriately based on vehicle
states [8]. Direct integration methods can accumulate
sensor errors and unwanted measurements from road
grade and bank angle. In addition, methods based on
a physical vehicle model can be sensitive to changes
in the vehicle parameters and are only reliable in the
linear region.

To overcome these drawbacks, a new method for

estimating vehicle sideslip angle using GPS and INS
sensor measurements is presented in [7]. In this
Figure 1: Bicycle model. scheme, GPS measurements from a two-antenna
system are combined with INS sensor measurements
The sideslip angle is defined by the difference to eliminate errors due to direct integration. Since
between vehicle heading, ψ, and the direction of both the vehicle heading and the direction of velocity
velocity, γ: are directly measured from a two-antenna GPS
receiver, the sideslip angle can be calculated using
β = γ −ψ (1) Equation (1). INS sensors are integrated with GPS
measurements to provide higher update rate estimates
In Figure 1, δ is the steering angle, ux and uy are the of the vehicle states and to handle periods of GPS
longitudinal and lateral components of the CG signal loss. This method is also independent of any
velocity, Fyf and Fyr are the lateral tire forces front and parameter uncertainties and changes because it is
rear, respectively, and αf and αr are the tire slip angles. based on purely kinematic relationships.
Assuming constant longitudinal velocity ux=V, the 15
state equation for the bicycle model can be written as:
sideslip angle (deg)

10
5

 β&  
−C f − Cr
−1+ (C r b −C f a
)  β  +  Cf

δ
0
-5
 =
mV mV 2
 
mV (2)
Cr b −C f a −C f a 2 −C r b 2 Cf a -10
 r&   I IV   r   I  -15
0 50 100 150 200
tim e (s)

I is the moment of inertia of the vehicle about its yaw 60

40
axis, m is the vehicle mass, a and b are distance of the
yaw rate (deg/s)

20
front and rear axles from the CG, and Cf and Cr are the 0
total front and rear cornering stiffness that relate -20

lateral tire force to slip angle: -40

-60
0 50 100 150 200
tim e (s)
Fyf = C f α f
(3) Figure 2: Sideslip and yaw rate estimation.
Fyr = C r α r
Experimental results from the GPS/INS integration are
The model is valid for tires operating in the linear plotted in Figure 2 on top of simulation results from
region and small slip angles. the bicycle model for both yaw rate and sideslip angle.
The similarity between estimated and simulated yaw
rates indicates that the bicycle model used in the
3 State Estimation comparison is valid and calibrated correctly. The fact
that the sideslip measurement is clean and correlates
The ability to obtain accurate information on the with the model makes it suitable for use as a feedback
vehicle states—yaw rate and sideslip angle—is crucial signal.

why the yaw rate....side slip is as explained

before ..that it enhances the driving experience..
 4 Full State Feedback Controller steering system and allows the use of a much smaller
actuator. A rotary position sensor measures the lower
A full state feedback control law for an active steering steering shaft angle, which is equal to the front wheel
vehicle is given by steer angle scaled by the steering ratio. An identical
sensor attached to the upper steering shaft measures
δ = K r r + K β β + K dδ d (4) the handwheel angle.

where δd is the driver commanded steer angle and δ is feedback motor

belt drive
the augmented angle. A physically intuitive way to
modify a vehicle’s handling characteristics is to define handwheel angle sensor steering actuator
a target front cornering stiffness as
pinion angle sensor

Cˆ f = C f (1 + η ) (5)
power assist unit
pinion

and the state feedback gains as

a (6) rack
K β = −η Kr = − η K d = (1 + η )
V

where η is the desired fractional change in the original Figure 3: Steer-by-wire schematic.
front cornering stiffness Cf. Substituting the feedback
law (4) into Equation (2) yields a state space equation The servomotor actuator specifications are chosen
of the same form as Equation (2) but with the new based on the maximum torque and speed necessary to
cornering stiffness Ĉf: steer the vehicle under typical driving conditions
including moderate emergency maneuvers. On

 β&   −CmV
  = 
ˆ −C
f r
−1+ (Cr b −Cˆ f a
mV 2
) β  +  Cˆ f
mV

δ d (7)
average, steering torque required at the handwheel
during normal driving ranges from 0 to 2 Nm, while
emergency maneuvers can demand up to 15 Nm of
C b −Cˆ a −Cˆ f a 2 −Cr b 2 Cˆ f a
 r&   r I f IV
  r   I  torque [9]. The actuator installed in the test vehicle
provides a maximum steering torque of 17.1 Nm with
Since a vehicle’s handling characteristics are heavily a maximum steer rate of 700 degrees per second.
influenced by tire cornering stiffness, the effect of this
modification is to make the vehicle either more The differential equation describing the steering
oversteering or understeering depending on the sign of system dynamics is as follows:
η. Clearly, there are many other ways to apply full
state feedback, but the physical motivation behind Jθ&& + bθ& + Fc sgn θ& + k aτ a = τ (8)
cornering stiffness adjustment makes clear through the
bicycle model exactly how the handling characteristics θ is the pinion angle, J is the total moment of inertia of
have been modified. Note that in this formulation, it the system, b is viscous damping, Fc represents
is not necessary to know the real cornering stiffness of coulomb friction, ka is a scale factor, τa is the tire self-
the front tire—only vehicle speed and weight aligning moment, and τ is the actuator torque.
distribution, which are relatively easy to measure—to
achieve the desired handling modification. The purpose of the steer-by-wire controller is to track
commanded steer angle with minimal error; the
control effort consists of three components:
5 Steer-by-Wire System
τ = τ feedback + τ feedforward + τ aligning (9)
A production model 1997 Chevrolet Corvette is
modified for full steer-by-wire capability by replacing
The proportional derivative (PD) feedback component
the steering shaft with a brushless DC servomotor
is given by
actuator. The stock hydraulic power assist unit and
rack and pinion mechanism in the test vehicle are
τ feedback = K p (θ d − θ ) + K d (θ&d − θ& ) (10)
retained as part of the steer-by-wire system, since the
incorporation of the power assist unit eliminates the
need for extensive modifications to the existing
where θd is the desired steer angle, Kp is the tire self-aligning moment. The total aligning moment
proportional feedback constant, and Kd is the is given by
derivative feedback constant. The feedback gains Kp
and Kd are selected to give a fast closed loop system τ a = (t p + t m )Fyf (α f ) (12)
response without oscillatory behavior. Because the
system is second order, however, PD control alone where tp and tm are the tire pneumatic and mechanical
results in some steady state error when tracking the trails, respectively. Front tire slip angle, αf, can be
type of command shown in Figure 4 (steering angle is calculated from the following relationship involving
given at the front wheels). To obtain these estimated sideslip and other measurable parameters:
measurements, the front wheels are raised off the
ground so as to isolate the influence of J, b and Fc ar (13)
from static friction at the tire-ground interface. The αf = β + −δ
ux
addition of feedforward compensation,
Aligning moment may also be directly approximated
τ feedforwar d = Jθ&&d + bθ&d + Fc sgn (θ&d ) (11) as an empirical function of tire slip angle [10]. This
approximation of aligning moment is added to the
to the PD controller cancels any tracking errors feedback and feedforward control as
associated with the system dynamics and internal
friction (Figure 5). J, b and Fc are determined through τ aligning = k aτˆa (α f ) (14)
closed-loop identification of the steering system.
10 where ka is a scale factor to account for torque
actual
reduction by the steering gear.
steering angle (deg)

5 com m anded

0 20
actual
steering angle (deg)

10 com m anded
-5

-10 0
6 8 10 12 14 16 18 20 22 24
tim e (s) -10
0.5
steering angle error (deg)

-20
10 15 20 25 30 35 40 45 50 55
tim e (s)

1
steering angle error (deg)

0.5

-0.5 0
6 8 10 12 14 16 18 20 22 24
tim e (s) -0.5

Figure 4: Feedback control only. -1

10 15 20 25 30 35 40 45 50 55
tim e (s)
10
actual
Figure 6: Error due to aligning moment.
steering angle (deg)

5 com m anded

0 20
actual
steering angle (deg)

10 com m anded
-5

-10 0
6 8 10 12 14 16 18 20 22 24
tim e (s) -10
0.5
steering angle error (deg)

-20
10 15 20 25 30 35 40 45 50 55
tim e (s)

1
steering angle error (deg)

0.5

-0.5 0
6 8 10 12 14 16 18 20 22 24
tim e (s) -0.5

Figure 5: Feedback with feedfoward compensation. -1

10 15 20 25 30 35 40 45 50 55
tim e (s)

When driving a vehicle over the road, however, an Figure 7: Steering controller with aligning moment
additional disturbance acts on the system causing a compensation.
steering error (Figure 6) that is directly attributable to
From a comparison between Figures 6 and 7, the rate and sideslip values than the nominal case. This
addition of τaligning to the actuator effort effectively behavior is expected since reducing the front
eliminates most of the steering disturbances that arise cornering stiffness causes the vehicle to tend toward
when turning at speed. understeer. Figure 11 confirms that test results for the
reduced case match bicycle model simulation.
40
6 Experimental Results norm al
reduced
30

20

10

yaw rate (deg/s)

0

-10

-20

-30

-40
5 10 15 20 25
tim e (s)

Figure 10: Comparison between normal and

Figure 8: Steer-by-wire test vehicle. effectively reduced front cornering stiffness.
40
The steer-by-wire test vehicle is equipped with sim ulation
experim ent
multiple-antenna GPS configured to provide absolute 30

velocity and heading information. INS sensors 20

measure lateral and longitudinal acceleration, yaw
rate, and roll rate. The experimental setup for vehicle 10
yaw rate (deg/s)

state estimation is same as described in [7]. In Figure 0

9, the measured yaw rate from a sinusoidal steering
input while driving at 13.4 m/s (30 mi/hr) compare -10

well to simulation results from the bicycle model. -20

40 -30
sim ulation
experim ent
30 -40
5 10 15 20 25
tim e (s)
20

Figure 11: Comparison between bicycle model and

10
experiment with reduced cornering stiffness.
yaw rate (deg/s)

-10
Experimental data show a corresponding but opposite
change in handling behavior when the effective front
-20 cornering stiffness is increased such that the vehicle
-30
tends toward oversteer.

-40
5 10 15 20 25 For the final series of tests, 182 kg (400 lbs) of weight
tim e (s)
are added to the rear of the vehicle so that 57% of the
Figure 9: Comparison between bicycle model and total vehicle weight lies over the rear axle with 43%
experiment with normal cornering stiffness. over the front axle. The unloaded vehicle has a
weight distribution balanced equally front to rear. As
Next, handling modification is implemented on the seen in Figure 12, the loaded vehicle exhibits slightly
test vehicle. Changes in handling behavior under full more oversteering behavior than the unloaded vehicle.
state feedback control are evaluated by comparing However, with active handling modification, a 20%
measured vehicle response to the nominal case shown reduction in front cornering stiffness returns the
in Figure 9. In Figure 10, the effective front cornering controlled vehicle to the near neutral handling
stiffness is reduced 50% by setting the parameter η to behavior of the unloaded vehicle (Figure 13). While
-0.5. The experimental results exhibit lower peak yaw the difference in handling behavior may seem small
when viewed on a graph, the improvement is readily fundamental limitations imposed by the feedback or
apparent to both driver and passenger. control structure.
40
unloaded
loaded
30 Acknowledgements
20
The authors wish to acknowledge Michael Grimaldi,
10 Robert Wiltse and Pamela Kneeland at General
yaw rate (deg/s)

0
Motors Corporation for their donation of the test
vehicle and the GM Foundation for the grant enabling
-10
its conversion to steer-by-wire. Thanks also to Dr.
-20 Skip Fletcher, T.J. Forsyth, Geary Tiffany and Dave
Brown at the NASA Ames Research Center for
-30
providing the use of Moffett Federal Airfield for
-40 vehicle test purposes.
5 10 15 20 25
tim e (s)

Figure 12: Comparison between unloaded and loaded

vehicle. References
40
unloaded
[1] J. Kasselmann and T. Keranen. Adaptive steering.
30
loaded, reduced Bendix Technical Journal, vol. 2, pp. 26-35, 1969.
[2] J. Ackermann. Yaw disturbance attenuation by robust
20 decoupling of car steering. In Proceedings of the IFAC
World Congress, San Francisco, CA, 1996.
10
[3] K. Huh and J. Kim. Active steering control based on
yaw rate (deg/s)

0
the estimated tire forces. Journal of Dynamic Systems,
Measurement, and Control, vol. 123, pp. 505-511, 2001.
-10 [4] M. Segawa, K. Nishizaki and S. Nakano. A study of
vehicle stability control by steer by wire system. In
-20
Proceedings of the International Symposium on Advanced
-30 Vehicle Control (AVEC), Ann Arbor, MI, 2002.
[5] M. Nagai, S. Yamanaka, and Y. Hirano. Integrated
-40
5 10 15 20 25 control law of active rear wheel steering and direct yaw
tim e (s)
moment control. In Proceedings of the International
Figure 13: Comparison between unloaded vehicle and Symposium on Advanced Vehicle Control (AVEC), Aachen,
loaded vehicle with handling modification. Germany, 1996.
[6] A. van Zanten. Evolution of electronic control systems
for improving the vehicle dynamic behavior. In
Proceedings of the International Symposium on Advanced
7 Conclusion Vehicle Control (AVEC), Tokyo, 2002.
[7] J. Ryu, E. Rossetter, and J.C. Gerdes. Vehicle sideslip
This work represents one of the first applications of and roll parameter estimation using GPS. In Proceedings of
GPS-based state estimation to dynamic control of a the International Symposium on Advanced Vehicle Control
vehicle with active steering. A full state feedback (AVEC), Tokyo, 2002.
controller has been developed to alter a vehicle’s [8] Y. Fukada. Estimation of vehicle slip-angle with
handling characteristics by augmenting the driver’s combination method of model observer and direct
integration. In Proceedings of the International Symposium
steering input. The controller is experimentally
on Advanced Vehicle control. (AVEC), Nagoya, Japan,
validated on a steer-by-wire vehicle equipped with 1998.
GPS and INS sensors. Experimental results confirm [9] A. Liu and S. Chang. Force feedback in a stationary
that it is possible to effectively change the cornering driving simulator. In Proceedings of the IEEE International
stiffness of the front tires by full state feedback Conference on Systems, Man, and Cybernetics, vol. 2,
modification of the driver’s steering command. Thus, pp.1711-1716, Vancouver, BC, 1995.
a vehicle’s handling characteristics may be tuned to [10] H. Pacejka, E. Bakker and L. Nyborg. Tyre modelling
driver preference or adjusted for variations in for use in vehicle dynamics studies. Society of Automotive
operating conditions such as load distribution. Future Engineers technical paper no. 870421.
work will investigate the possible extent of vehicle
handling modification by active steering and any