Visit https://ebooknice.

com to download the full version and

explore more ebooks

(Ebook) Synchronization and Control of Multiagent

Systems (Automation and Control Engineering) by Dong
Sun ISBN 9781439820476, 9781439820483, 1439820473,
1439820481
_____ Click the link below to download _____
https://ebooknice.com/product/synchronization-and-
control-of-multiagent-systems-automation-and-control-
engineering-1657670

Explore and download more ebooks at ebooknice.com

Here are some recommended products that might interest you.
You can download now and explore!

(Ebook) Biota Grow 2C gather 2C cook by Loucas, Jason; Viles, James

ISBN 9781459699816, 9781743365571, 9781925268492, 1459699815,
1743365578, 1925268497

https://ebooknice.com/product/biota-grow-2c-gather-2c-cook-6661374

ebooknice.com

(Ebook) SAT II Success MATH 1C and 2C 2002 (Peterson's SAT II Success)

by Peterson's ISBN 9780768906677, 0768906679

https://ebooknice.com/product/sat-ii-success-
math-1c-and-2c-2002-peterson-s-sat-ii-success-1722018

ebooknice.com

(Ebook) Matematik 5000+ Kurs 2c Lärobok by Lena Alfredsson, Hans

Heikne, Sanna Bodemyr ISBN 9789127456600, 9127456609

https://ebooknice.com/product/matematik-5000-kurs-2c-larobok-23848312

ebooknice.com

(Ebook) Master SAT II Math 1c and 2c 4th ed (Arco Master the SAT
Subject Test: Math Levels 1 & 2) by Arco ISBN 9780768923049,
0768923042

https://ebooknice.com/product/master-sat-ii-math-1c-and-2c-4th-ed-
arco-master-the-sat-subject-test-math-levels-1-2-2326094

ebooknice.com
(Ebook) Cambridge IGCSE and O Level History Workbook 2C - Depth Study:
the United States, 1919-41 2nd Edition by Benjamin Harrison ISBN
9781398375147, 9781398375048, 1398375144, 1398375047

https://ebooknice.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-53538044

ebooknice.com

(Ebook) Multiagent Systems: Introduction and Coordination Control by

Magdi S. Mahmoud ISBN 9780367255367, 0367255367

https://ebooknice.com/product/multiagent-systems-introduction-and-
coordination-control-11290482

ebooknice.com

(Ebook) Automatic Control, Mechatronics and Industrial Engineering:

Proceedings of the International Conference on Automatic Control,
Mechatronics and Industrial Engineering (ACMIE 2018), October 29-31,
2018, Suzhou, China by Yigang He (Editor); Xue Qing (Editor) ISBN
9780429468605, 9780429888175, 9780429888199, 9781138604278,
https://ebooknice.com/product/automatic-control-mechatronics-and-
0429468601, 0429888171, 0429888198, 1138604275
industrial-engineering-proceedings-of-the-international-conference-on-
automatic-control-mechatronics-and-industrial-engineering-
acmie-2018-october-29-31-2018-suzhou-china-12057506
ebooknice.com

(Ebook) Impulsive Synchronization of Complex Dynamical Networks:

Modeling, Control and Simulations by Ze Tang, Dong Ding, Yan Wang,
Zhicheng Ji, Ju H. Park ISBN 9789811653827, 9811653828

https://ebooknice.com/product/impulsive-synchronization-of-complex-
dynamical-networks-modeling-control-and-simulations-34493448

ebooknice.com

(Ebook) Modeling and Control of Complex Systems (Automation and

Control Engineering) by Petros A. Ioannou, Andreas Pitsillides ISBN
0849379857

https://ebooknice.com/product/modeling-and-control-of-complex-systems-
automation-and-control-engineering-2120528

ebooknice.com
 Synchronization
and Control of
Multiagent Systems
 AUTOMATION AND CONTROL ENGINEERING
A Series of Reference Books and Textbooks

Series Editors
FRANK L. LEWIS, Ph.D., SHUZHI SAM GE, Ph.D.,
Fellow IEEE, Fellow IFAC Fellow IEEE
3URIHVVRU 3URIHVVRU
$XWRPDWLRQ DQG 5RERWLFV 5HVHDUFK ,QVWLWXWH ,QWHUDFWLYH 'LJLWDO 0HGLD ,QVWLWXWH
7KH 8QLYHUVLW\ RI 7H[DV DW $UOLQJWRQ 7KH 1DWLRQDO 8QLYHUVLW\ RI 6LQJDSRUH

Synchronization and Control of Multiagent Systems, Dong Sun

Subspace Learning of Neural Networks, Jian Cheng Lv, Zhang Yi, and Jiliu Zhou
Reliable Control and Filtering of Linear Systems with Adaptive Mechanisms,
Guang-Hong Yang and Dan Ye
Reinforcement Learning and Dynamic Programming Using Function
Approximators, Lucian Buşoniu, Robert Babuška, Bart De Schutter,
and Damien Ernst
Modeling and Control of Vibration in Mechanical Systems, Chunling Du
and Lihua Xie
Analysis and Synthesis of Fuzzy Control Systems: A Model-Based Approach,
Gang Feng
Lyapunov-Based Control of Robotic Systems, Aman Behal, Warren Dixon,
Darren M. Dawson, and Bin Xian
System Modeling and Control with Resource-Oriented Petri Nets, Naiqi Wu
and MengChu Zhou
Sliding Mode Control in Electro-Mechanical Systems, Second Edition,
Vadim Utkin, Jürgen Guldner, and Jingxin Shi
Optimal Control: Weakly Coupled Systems and Applications, Zoran Gajić,
Myo-Taeg Lim, Dobrila Skatarić, Wu-Chung Su, and Vojislav Kecman
Intelligent Systems: Modeling, Optimization, and Control, Yung C. Shin
and Chengying Xu
Optimal and Robust Estimation: With an Introduction to Stochastic Control
Theory, Second Edition, Frank L. Lewis, Lihua Xie, and Dan Popa
Feedback Control of Dynamic Bipedal Robot Locomotion, Eric R. Westervelt,
Jessy W. Grizzle, Christine Chevallereau, Jun Ho Choi, and Benjamin Morris
Intelligent Freight Transportation, edited by Petros A. Ioannou
Modeling and Control of Complex Systems, edited by Petros A. Ioannou
and Andreas Pitsillides
Wireless Ad Hoc and Sensor Networks: Protocols, Performance, and Control,
Jagannathan Sarangapani
Stochastic Hybrid Systems, edited by Christos G. Cassandras
and John Lygeros
Hard Disk Drive: Mechatronics and Control, Abdullah Al Mamun,
Guo Xiao Guo, and Chao Bi
Autonomous Mobile Robots: Sensing, Control, Decision Making
and Applications, edited by Shuzhi Sam Ge and Frank L. Lewis
Automation and Control Engineering Series

Synchronization
and Control of
Multiagent Systems

Dong Sun
City University of Hong Kong
Kowloon, Hong Kong, People’s Republic of China

Boca Raton London New York

CRC Press is an imprint of the

Taylor & Francis Group, an informa business
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2011 by Taylor and Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1

International Standard Book Number: 978-1-4398-2047-6 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Sun, Dong, 1967-

Synchronization and control of multiagent systems / Dong Sun.
p. cm. -- (Automation and control engineering)
Includes bibliographical references and index.
ISBN 978-1-4398-2047-6 (hardback)
1. Automatic control. 2. Multiagent systems. 3. Robotics. 4. Synchronization. I. Title.

TJ213.S7985 2010
629.8--dc22 2010035470

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com
 Dedication

To My Family—Jiang, Sheryl, and My Parents

Contents
Preface.......................................................................................................................xi
Acknowledgments................................................................................................... xiii
About the Author...................................................................................................... xv

Chapter 1. Introduction...........................................................................................1
1.1 Background.................................................................................1
1.2 Synchronization..........................................................................7
1.3 Outline of the Book.................................................................. 10
1.4 Summary.................................................................................. 11
References........................................................................................... 12

Chapter 2. Synchronization Strategy.................................................................... 17

2.1 Concept of Synchronization..................................................... 17
2.2 Synchronization Control Goal..................................................20
2.3 Synchronization Errors.............................................................25
2.4 Summary.................................................................................. 27
References...........................................................................................28

Chapter 3. Model-Free Synchronization Control of Multiple Motion Axes......... 29

3.1 Problem Statement.................................................................... 29
3.2 Position Synchronization Errors and Control Strategy............ 32
3.3 Multiaxis Synchronization in Setpoint Position Control.......... 33
3.4 Multiaxis Synchronization in Tracking Control....................... 37
3.5 Experiments..............................................................................40
3.6 Summary.................................................................................. 48
References........................................................................................... 52

Chapter 4. Synchronized Control of Multiaxis Systems in Trajectory

Tracking.............................................................................................. 55
4.1 Synchronization Strategy of Multiagent Motions.................... 55
4.1.1 Synchronization Strategy............................................ 57
4.2 Model-Based Cross-Coupling Synchronization Control.......... 57
4.3 Adaptive Synchronization Control........................................... 62
4.4 Case Study—Adaptive Coupling Control of Two
Working Operations in Computer Numerical Control
Integrated Machine...................................................................66
4.5 Summary.................................................................................. 71
References........................................................................................... 71

vii
viii Contents

Chapter 5. Adaptive Synchronization Control for Coordination

of Multiple Robot Manipulators.......................................................... 73
5.1 Introduction.............................................................................. 73
5.2 Motion Synchronization Strategy of Multiple
Manipulators............................................................................. 75
5.3 Adaptive Synchronization Control........................................... 78
5.4 Case Studies............................................................................. 83
5.4.1 Experiments of Coordinating Two Industrial
Manipulators............................................................... 83
5.4.2 Simulations of Coordinating Multiple
Manipulators............................................................... 89
5.5 Summary..................................................................................97
References........................................................................................... 98

Chapter 6. Synchronization Control for Minimization of Contouring

Errors of Computer Numerically Controlled Machine Tools........... 101
6.1 Introduction............................................................................ 102
6.2 Modeling of a Computer Numerical Control
Machine Tool.......................................................................... 104
6.3 Contouring Errors and Synchronization Errors..................... 105
6.3.1 Contouring Errors..................................................... 105
6.3.2 Synchronization Errors............................................. 106
6.4 Control Design........................................................................ 108
6.4.1 Controller Formulation.............................................. 108
6.4.2 Stability Analysis...................................................... 110
6.5 Experiments............................................................................ 113
6.6 Summary................................................................................ 122
References......................................................................................... 125

Chapter 7. Synchronization Control of Parallel Robotic Manipulators.............. 127

7.1 Introduction............................................................................ 127
7.2 Modeling and Synchronization Error of Parallel
Manipulators........................................................................... 130
7.2.1 Modeling................................................................... 130
7.2.2 Synchronization Error............................................... 131
7.3 Control Design........................................................................ 133
7.4 Experiments............................................................................ 138
7.5 Summary................................................................................ 142
References......................................................................................... 144

Chapter 8. A Synchronization Approach to Multirobot Formations.................. 147

8.1 Introduction............................................................................ 147
8.2 Multirobot Formation via Synchronization............................ 150
Contents ix

8.3 Control Design........................................................................ 153

8.3.1 Synchronous Formation Controller........................... 154
8.3.2 Discussions................................................................ 158
8.3.2.1 Boundedness.............................................. 158
8.3.2.2 Adaptive Control for Robustness............... 159
8.4 Simulations............................................................................. 160
8.5 Multirobot Formation Experiments........................................ 166
8.5.1 Case 1: Triangle Formation....................................... 166
8.5.2 Case 2: Ellipse Formation......................................... 168
8.6 Summary................................................................................ 174
Appendix........................................................................................... 174
References......................................................................................... 175
Index....................................................................................................................... 179
Preface
Rapid advances in sensing, computing, and communication technologies have led to
the development of autonomous systems functioning individually in uncertain envi-
ronments, opening up new challenges to understanding and developing cooperative
multiagent systems. There are many examples of multiagent systems, such as multi­
robot cooperation and multiaxis computer numerical control (CNC) machining, in
which multiple agents work cooperatively to achieve a common goal. Multiagents
are envisaged to help accomplish tasks that could not possibly be completed with
individual agents acting alone. A higher success rate and a high level of reliability in
the handling of tasks can be achieved if multiple agents are endowed with coopera-
tion capabilities.
Synchronization control provides unique advantage and opportunity to solve the
problem of multiagent coordination. Utilizing a cross-coupling concept, a synchro-
nization control framework can be developed such that all agents accomplish each
individual task while synchronizing motions among them to maintain relative kine-
matics relationships to meet the coordination requirement. The synchronization con-
trol is ideally suited to multiagent systems in performing a group task as a whole.
Some examples are multirobot assembly, multiaxis CNC machining, and formation
control of networked robots.
This book will give a detailed introduction of the cross coupling–based synchroni-
zation control approach to multiagent systems. The following chapters are included:

Chapter 1: Introduction
Chapter 2: Synchronization Strategy
Chapter 3: Model-Free Synchronization Control of Multiple Motion Axes
Chapter 4: Synchronized Control of Multiaxis Systems in Trajectory Tracking
Chapter 5: Adaptive Synchronization Control for Coordination of Multiple
Robot Manipulators
Chapter 6: Synchronization Control for Minimization of Contouring Errors of
Computer Numerically Controlled Machine Tools
Chapter 7: Synchronization Control of Parallel Robotic Manipulators
Chapter 8: A Synchronization Approach to Multirobot Formations

The key objectives of this book are as follows:

Building a connection between the multiagent coordinate task and the synchro-
nization approach. It will be shown in this book how to pose the multiagent
control problem as a synchronization control problem, permitting each
agent to be part of the coordination system while recognizing its individual
task performance capability.
Developing a theoretical framework and methodology for cooperation among
multiple agents, capable of addressing the problems of uncertain dynamic

xi
xii Preface

models and unknown environmental disturbances. A synchronous coordi-

nation control methodology will be reported, which can guarantee both
position tracking and synchronization errors to converge to zero.
Performing application studies to demonstrate the effectiveness of the pro-
posed synchronization approach. Simulations and experiments will be
performed on various multiagent systems, such as a multiaxis CNC
machine, multiple robot manipulator, parallel manipulators, and multi-
robot in formation.

Applications of synchronization control can be found in three main areas: manu-

facturing industry, civil applications, and system biology and human health. In the
manufacturing industry, synchronization control can be widely applied to various
manufacturing automation tasks supported by surface mounting technology (SMT)
devices, CNC machines, and multirobot work cells. In civil applications, synchroni-
zation control can be used for search and rescue operations utilizing vehicles such as
helicopters or ships, and control of traffic jams in intelligent transportation systems.
Synchronization control also exhibits great potential in many applications of system
biology and human health, although this is not fully addressed in this edition of the
book. Examples include simulating and studying human interaction aspects, such as
the spread of diseases (life science), and synchronous control of multiple biological
cells in microscale or nanoscale in cell manipulation.
This book will be one of the first to systematically introduce the knowledge of
synchronization controls for multiagent systems. In addition to detailed theoretical
approaches, the book will give numerous application examples. It can be used as a
textbook for university students or a handbook for engineers to solve practical engi-
neering problems.
Acknowledgments
Several people should be mentioned for their contributions to this book. First, I
would like to thank Professor James K. Mills of the University of Toronto, Canada,
for his longtime coordination and support of this project. Second, I would like to
thank the following research associates and students at the City University of Hong
Kong and the University of Toronto for their assistance during the production of this
book: Chong Liu, Can Wang, Xiaoyin Shao, Ming Chau Tong, Chi Ming Lam, and
Lu Ren. My thanks also go to my colleague, Professor Gang Feng, who provided
valuable insights to both project implementation and the book-writing process. I am
also grateful for the help provided by the editor Li Ming Leong at Taylor & Francis
Asia Pacific.
Finally, I would like to thank Professor Sam Shuzhi Ge of the University of
Singapore and Professor Frank Lewis of the University of Texas at Arlington, for
their kind invitation and encouragement in writing this book.

xiii
About the Author
Dong Sun received BSc and MSc degrees in mechatronics and biomedical engineer-
ing from Tsinghua University, China, and a PhD degree in robotics and automation
from the Chinese University of Hong Kong. From 1997 to 1999, he worked with
the University of Toronto as a postdoctoral researcher. After a short time working
as a research and development (R&D) engineer in industry in Ontario, Canada, he
joined the department of Manufacturing Engineering and Engineering Management
at the City University of Hong Kong in 2000, where he is now a full professor. He
also holds an adjunct faculty position at the University of Toronto, and is a licensed
professional engineer in Ontario, Canada.
Sun’s research interests lie in multirobot systems, robotics manipulation, motion
control, and biological processing automation. Since 2000, he has obtained more
than 30 million dollars in research grants as the principal investigator and led more
than 20 research projects toward successful completion. He has published 200 tech-
nical articles in refereed journals and conference proceedings and held four patents.
He is a fellow of the Hong Kong Institution of Engineers.
He has received numerous awards, including the Best Paper award of the 2008
Institute of Electrical and Electronics Engineers (IEEE) International Conference
on Robotics and Biomimetics, 2007 Outstanding Paper Award of IEEE Transactions
on Fuzzy System, Gold Award of 2006 Hong Kong Electronics Fair, 2004 Applied
Research Excellence Award (Certificate of Merit) of City University of Hong Kong,
and 2003 Hong Kong Award for Industry.
Sun has been actively involved in various professional activities. He serves as
an associate editor for IEEE Transactions on Robotics, technical editor for IEEE/
ASME Transactions on Mechatronics, and associate editor for the IEEE Robotics
and Automation Society Conference Board. He was chairman of the IEEE Hong
Kong section joint chapter of Robotics and Automation and Control Systems in 2007
and 2008, and led the chapter to win the best chapter award of the IEEE Robotics
and Automation Society in 2008. He has been a Program Committee and Organizing
Committee member for many international conferences, including serving as the pro-
gram chair of the 2009 IEEE International Conference on Robotics and Biomimetics,
and the general chair of the 2010 IEEE International Conference on Nano/Molecular
Medicine and Engineering.

xv
 1 Introduction

Abstract
This chapter will present a broad overview of multiagent systems in various
applications and a general introduction of synchronization controls, along with
basic concepts necessary to ensure understanding of the topic.
In Section 1.1, the reader will have an opportunity to learn about the back-
ground of multiagent systems and their coordination as well as future promises,
and a summary of numerous challenging problems encountered by multiagent
coordination will be presented. In Section 1.2, the reader will be exposed to
a broad range of concepts and practices of various synchronization control
technologies to multiagent systems in manufacturing automation and robotics,
where the cross-coupling control approach plays a promising role in synchro-
nizing the motions of multiple agents. Section 1.3 will present an outline of
this book. A short summary will be given in Section 1.4.

1.1 Background
Rapid advances in sensing, computing, and communication technologies have led
to the development of autonomous systems functioning individually in uncertain
environments, opening new challenges to understanding and developing cooperative
multiagent systems. Multiple agents are envisaged to help accomplish tasks that can-
not be completed with individual ones acting alone. A higher success rate and a high
level of reliability in handling those complex tasks can be achieved if multiple agents
are endowed with cooperation capabilities. Some examples of multiagent systems
include multirobot cooperation and multiaxis computer numerical control (CNC)
machining, in which multiple agents work cooperatively to achieve a common goal.
Nowadays, studies of multiagent systems and their cooperative controls have
become a popular research area with dramatically increased popularity. The idea of
multiagent working and cooperation was inspired by many examples in biology and
life science. Figure 1.1 illustrates a few biological examples of multiagent systems in
nature: ant swarming, bird flocking, and fish schooling. It is well known that multi-
agent systems, if working cooperatively under highly efficient organizations and prin-
ciples, can behave as a whole, guaranteed with fault tolerance and robust properties.
Studies of multiagent coordinative controls have attracted considerable attention
in past decades. Of these studies, two important developmental stages are gener-
ally recognized. The first stage was in the 1980s and 1990s. Many studies in this
stage focused on the coordination of multiple robot manipulators or CNC machin-
ing axes for industrial applications. The second stage started around the year 2000.

1
2 Synchronization and Control of Multiagent Systems

(a) (b)

(c)

Figure 1.1 Biological examples of multiagent systems in nature: (a) ant swarming, (b) bird
flocking, and (c) fish schooling.

More studies in this stage focused on large scalability of the coordinated agents (i.e.,
formation and consensus controls of swarm mobile agents, in which technologies
regarding multiagent networks attract increasing attention).
The motivation behind the first-stage studies was high demand for coordination
of multiple robots in carrying out assembly tasks in both industrial and space appli-
cations. Figure 1.2 illustrates a typical example of coordinating multiple manipula-
tors in the automobile manufacturing industry. Three major coordination schemes
aiming for hybrid position and force controls have been reported in the literature.
The first scheme is the master/slave control (Arimoto et al. 1987; Luh and Zheng
1987), where one robot arm is under position control, and the others are subject
to compliant force control to maintain kinematic constraints. The second scheme
utilizes centralized control architecture (Koivo and Unseren 1991; Tarn et al. 1986;
Wen and Delgado 1992; Yoshikawa et al. 1988; Yun et al. 1997), in which robots
and the grasped payload are considered as a closed kinematic chain. This method is
designed based on a unified robot and payload dynamic model. The third scheme is a
Introduction 3

Figure 1.2 Automobile manufacturing with multirobot work cell.

decentralized control (Hsu 1989; Kosuge et al. 1993; Liu et al. 1996; Xi et al. 1996),
in which each robot is controlled separately by its own local controller.
Since the late 1990s, second-stage studies on multiagent systems with larger scal-
ability have become more necessary than ever before to successfully execute various
tasks in a dynamic and uncertain environment. A typical application is that swarms
of mobile robots engaged in multirobot tasks must alter their formation according
to the changing environment, and in response to unexpected changes in the task
requirements. Relevant applications include exploration (Fox et al. 2000), coopera-
tive robot reconnaissance (Balch and Arkin 1998) and manipulation (Tanner et al.
2003), formation flight control (Mesbahi and Hadaegh 2001), satellite clustering
(McInnes 1995), and control of groups of unmanned air vehicles (Giulietti et al.
2000; Stilwell and Bishop 2000). A common characteristic in these applications is
that all the agents form a networked system to execute tasks in a coordinated way.
Figure 1.3 illustrates examples of practical applications of multiagent networked sys-
tems. Among the interests, one popular study is the formation control of swarms
of mobile robots, which have attracted considerable attention in recent years. At
present, the formation controls mainly include behavior-based control (Balch and
Arkin 1998; Berman et al. 2003; Lawton et al. 2003; Parker 1998), virtual structure
approach (Beard et al. 2001; Egerstedt and Hu 2001; Kang et al. 2000; Lewis and
Tan 1997), and leader–follower strategy (Chen et al. 2010; Das et al. 2002; Deasi
et al. 2001; Huang et al. 2006). The other approaches include artificial potential–
based methods (Ogren et al. 2004; Sepulchre et al. 2007) and graph theory–based
4 Synchronization and Control of Multiagent Systems

(a) (b)

(c)

Figure 1.3 Examples of practical applications of networked multiagent systems: (a) air
fleet, (b) tank team, and (c) robot carriers.

methods (Belta and Kumar 2004; Fax and Murray 2004; Jadbabaie 2003; Moreau
2005; Olfati-Saber 2006; Olfati-Saber and Murray 2004; Ren and Beard 2005). It
should be pointed out that much ongoing research in this area is still in the beginning
stages, and the delivered results are not satisfactory enough to address the challenges
presented by many practical applications.
More recently, the technology of multiagent systems at the microscale and nano-
scale has exhibited great potential in biological applications. In the last few years, there
has been growing interest in the development of methodologies for robotics manipula-
tion of microscaled biological cells. During this biological manipulation, people nor-
mally work with a group of cells or even an organ. If each cell is treated as a single
agent, manipulation of a group of multiple cells is a cooperative work of networked
agents. Figure 1.4 illustrates an example of multiple primary T cells in a three-dimen-
sional extracellular matrix (Sowinski et al. 2008), from which the membrane nanotubes
formed among them can be seen. In order to analyze the nanotubes, the cells connected
with nanotubes must be separated from those without nanotubes through microma-
nipulation. In moving the cells connected with nanotubes, motions of the cells must
be coordinated. Otherwise, nanotubes will be broken. Figure 1.5 illustrates another
Introduction 5

Figure 1.4 Multiple primary T cells and the membrane nanotubes formed among them.
(From Sowinski, S., Jolly, C., Berninghausen, O., Purbhoo, M. A., Chauveau, A., Köhler, K.,
Oddos, S., Eissmann, P., Brodsky, F. M., Hopkins, C., Önfelt, B., Sattentau, Q., and Davis,
D. M. 2008. Membrane nanotubes physically connect T cells over long distances presenting
a novel route for HIV-1 transmission. Nature Cell Biology 10(2):211–219.)

example of tissue organization, culture, and analysis in microsystems (El-Ali et al.

2006). Microsystems create new opportunities for the spatial and temporal control of
cell growth and stimuli. Further integration with bioanalytic microsystems results in a
multifunctional platform for basic biological insights into cells and tissues, as well as
for cell-based sensors with biochemical, biomedical, and environmental functions.
As seen from the above, the coordination of multiagent systems has endured a
rapid and dynamic growth worldwide over the last two decades. Government and
private agencies in many countries have funded and supported these research activi-
ties for both civil and military perspectives. It is expected that the relevant research

Cell imaging
Tissue organization Cell selection

Growth media

Simulation Cell lysis

Biochemical
analysis

Figure 1.5 Tissue organization, culture, and analysis in microsystems. (From El-Ali,
Sorger, J. P. K., and Jensen, K. F. 2006. Cells on chips. Nature 442(7101):403–411.)
6 Synchronization and Control of Multiagent Systems

will continue to grow rapidly in the next decades. Advancements of multiagent con-
trols will likely be manifested in several aspects: (1) an increased functional reach
of multiagent systems in interdisciplinary applications; (2) the maturation of vari-
ous coordination technologies; (3) the enhancement of system performances such
as functional quality, effectiveness, and robustness; (4) the seamless integration of
controls, sensing, and intelligence and network technologies in a functional system;
and (5) a lowered development cost and cost of ownership.
Coordination of multiagent systems encounters numerous challenges:

1. In a multirobot assembly system, the robots are not physically connected

but are required to perform a common task that must be achieved by coor-
dinating all the robots.
This challenging problem exists in many industrial applications (e.g., one
robot manipulates a payload while another spreads adhesive on the edges in a
robot assembly work cell, or multiple motion axes of a surface mounting tech-
nology [SMT] machine work coordinately in printed circuit board [PCB] manu­
facturing). Conventional centralized and decentralized coordination schemes
such as the hybrid position and force control algorithm based on a unified mul-
tirobot model where all the robots are assumed to be physically connected,
have not addressed this kind of coordination problem. The master/slave control
scheme may be realized as a solution to such coordination, but the performance
of the master/slave control may be adversely affected due to a communication
delay among robots. Moreover, complicated setup and implementation of the
hybrid position and force control architecture prevent the coordination schemes
from being widely used for many commercially available robots. There exists
a significant demand for easy and efficient implementation of the coordination
algorithm to meet the needs of practical applications.
2. In the formation control of networked robots, all the robots maintain in a
time-varying formation while carrying out a group task as a whole.
The organization of the grouped robots into a specific geometric shape
is generally believed to be the most efficient way to accomplish multiro-
bot cooperation tasks. For example, robotic observers marching toward the
same direction in specified formation geometry can maximize the search
area; and autonomous unmanned helicopters can reduce the group energy
cost if the group utilizes triangular formations in flying. A challenging
problem is how to ensure the robots maintain the required formation while
performing the group tasks, with rigorous proof of asymptotic stability to
show that both the trajectory tracking error and the formation error of each
robot converge to zero. The problem becomes extremely challenging when
the required formation geometry is time varying.
3. More and more multiagent systems demand capability in handling the prob-
lems of unknown dynamic models in a dynamic environment.
It is necessary to provide the multiagent system with the capability to
deal with various uncertain problems of the functional agents as well as
the dynamic environment that is hard to predict in advance. Vision technol-
ogy will play a key role in enabling the system to possess the function of
Introduction 7

intelligence, such as task recognition, online path and motion planning, and
simultaneous localization and mapping (SLAM). It is important to develop
scalable and adaptive solutions for intelligent coordination of multiagent
systems with uncertain dynamics in system models and environmental
disturbances. Note that in a tightly coupled multiagent system, each agent
should also adapt to the uncertainties of the other agents in addition to the
uncertainty of its own model.

This book will introduce how to utilize the synchronization control theory, a
novel and interesting technology to deal with multiple agents, in addressing these
challenging problems.

1.2 Synchronization
Synchronization is timekeeping that requires the coordination of events to operate a
system in unison. Synchronization is everywhere in our lives for different purposes
of safety, precision, efficiency, entertainment, and communication. This book will
emphasize how to synchronize motions of multiple agents.
One of the earliest works in synchronization of multiagent motions was made by
Koren (1980) in developing a cross-coupled controller. In this work, Koren took the
“equal-status” viewpoint to address the contouring performance in a two-axis feed
drive system. In the majority of traditional motion controls for multiagent systems,
the control loop of each agent receives local feedback information from this agent
only and no feedback from the other agents. Hence, disturbances in the control loop
of one agent cause an error that is corrected by this loop only, while the others do
not respond. Because motions of multiple agents must be coordinated to achieve
required objectives, any disturbance among the agents will subsequently degrade
the coordination performance. The cross coupling–based synchronization aims to
enable the control loop of each agent to receive feedbacks from itself as well as the
others for better coordination among the agents. Figure 1.6 illustrates the block dia-
gram of basic synchronization architecture using the cross-coupling concept.
It has been widely recognized that the cross-coupling control provides unique
advantages and opportunities to solve the synchronized control problem. Earlier stud-
ies of the synchronization control paid more attention to the coordination of multiple

Synchronization

…… Agent i–1 Agent i Agent i+1 ……

…… Controller Controller Controller ……

i–1 i i+1

Figure 1.6 Block diagram of cross-coupling synchronization architecture.

8 Synchronization and Control of Multiagent Systems

motion axes, partly due to its many applications to CNC machining. Kulkarni and
Srinivasan (1985, 1989) investigated the cross-coupled compensator scheme for con-
touring control of multiaxis machine tools, and further reported their cross-coupled
controls for biaxial feed drive servomechanisms in Kulkarni and Srinivasan (1990).
Tomizuka et al. (1992) incorporated the cross-coupled compensator to an adaptive
feedforward scheme to improve transient response and disturbance rejection, which
was followed by the works of Kamano et al. (1993) and Yang and Chang (1996).
These early approaches to synchronization control suffer from a limitation that
the multiaxis coordination goal must be of linear relationship among motion coor-
dinates of multiple axes. With the assumption that the desired contour was a com-
bination of piecewise linear segments, and the resulting controller was similar to
gain scheduling, Srinivasan and Fosdick (1988) proposed a multivariable analysis
approach to motion coordination. Koren and Lo (1991, 1992a, 1992b) introduced a
variable-gain cross-coupling controller for a general class of contours, neglecting the
effect of a time-varying gain in the cross-coupling controller to the system stability
and the effect of time-varying cross-coupling action to the overall system dynamics.
In 1992, Chiu and Tomizuka (1992) introduced a synchronization control algorithm
by taking into account the nonlinear coordination objectives (namely, nonlinear con-
tour) in the synthesis of the control algorithm.
Much of the reported research in the 1980s and early 1990s mainly dealt with the
velocity synchronization problem for synchronization of two-axis motions. Few of
these studies could address the position synchronization problem for multiple axes.
To solve this challenging problem, Chiu and Tomizuka (1998) applied the technique
of integrator backstepping in the synthesis of a time-domain control law. Sun (2003a)
proposed an adaptive synchronization control algorithm by incorporating the cross-
coupling approach to adaptive control architecture. In Sun (2003a), for the first time
it was shown that both position and synchronization errors of multiple motion axes
could converge to zero simultaneously, with the proposed adaptive synchronization
control. This method was experimentally verified in a case study of CNC industrial
application (Sun 2003b). The synchronization controls used for addressing the con-
tour tracking problem include the use of implicit representation of curves to model
desired contours (Chiu and Tomizuka 1998) and a receding horizon linear quadratic
formulation (McNab and Tsao 1994) and optimal control (McNab and Tsao 2000).
In practical applications, there exists a significant demand for not heavily using
dynamics models when implementing the synchronization controller. Koren and Lo
(1991) introduced a non-model-based variable-gain cross-coupling controller for
a general class of contours. This research, however, has yet to examine the effect
of a time-varying gain in the cross-coupling controller on system stability, and the
effect of the introduction of time-varying cross-coupling controls on the overall sys-
tem dynamics (Chiu and Tomizuka 2001). A control law that consists of a linear
time-varying proportional-derivative (PD) error feedback and a linear time invariant
trajectory feedforward compensator was proposed by Chiu and Tomizuka (2001),
in which the feedforward compensator was calculated from the model dynamics.
Sun et al. (2007) proposed a PD-type synchronization controller with feedback of a
properly defined coupled position error for a multiaxis motion control system, and
proved that such a controller could guarantee asymptotic convergence to zero of both
Introduction 9

position and synchronization errors in a setpoint position control. A setpoint tracking

controller was further developed by adding feedforward control terms and a satura-
tion function to the PD synchronization controller in Sun et al. (2007). Recently, Sun
and Tong (2009) reported a method of using a synchronization approach to minimize
contouring errors of a CNC machine tool, with a controller that did not require sig-
nificant use of the system dynamics models. Some other non-model-based synchro-
nization control approaches include fuzzy logic coupling control (Moore and Chen
1995) and neurocontroller (Lee and Jeon 1998), to name a few. An effort to examine
the stability and robustness of the cross-coupled control system was reported in Yeh
and Hsu (2003).
Applications of the synchronization approach to robotics started from motion
coordination of driven wheels in control of mobile robots. Borenstein and Koren
(1987) and Feng et al. (1993) applied a cross-coupling controller to synchronize
motions of two driven wheels of a differential mobile robot in trajectory tracking.
Borenstein (1995) proposed a concept of compliant linkage, which provided com-
pliance between the drive wheels of a vehicle, to accommodate control errors that
would otherwise cause wheel slippage. Sun et al. (2005) developed an orientation
control scheme by synchronizing two driven wheels of a differential mobile robot.
Further, the synchronization control was used for coordinative controls of mul-
tiple robot manipulators. A coordination of two robot arms with the cross-coupling
motion controller was reported in Naumovic (1999). Sun and Mills (2002) proposed
an adaptive synchronization controller to coordinate motions of multiple manipu-
lators in assembly tasks. In this work, the synchronization errors were defined as
differential position errors of end-effectors between every pair of two neighboring
manipulators, and both position and synchronization errors of end-effectors were
proven to converge to zero simultaneously under the proposed adaptive synchro-
nization control. Rodriguez-Angeles and Nijmeijer (2004) reported their work on
mutual synchronization of robot manipulators via estimated state feedback. Chung
and Slotine (2009) presented a synchronization tracking control law for cooperation
of multirobot systems and oscillation synchronization in robotic manipulation and
locomotion. The work also showed that complex dynamic networks could be con-
structed by exploiting concurrent synchronization such that multiple groups of fully
synchronized elements coexisted.
The importance of the synchronization control was found in control applications of
parallel manipulators. Because the trajectory of the parallel manipulator end-effector
is determined by motions of all actuators in the parallel manipulator, all actuated
joints can be controlled in a synchronous manner to achieve high tracking accuracy
in following a predetermined trajectory. Otherwise, the tracking accuracy may be
degraded due to uncoordinated motions of the actuated joints. In some severe situa-
tions, for instance, when accelerations of the actuated joints are high, damage to the
mechanical structure of the manipulator may occur. Chuang and Chang (2000) pro-
posed a contour error compensator based on the strategy of a Cartesian space cross-
coupled control and the transform relations between Cartesian space and joint space.
Sun et al. (2006) developed a decentralized trajectory tracking controller for a P-R-R
manipulator, where P and R represent prismatic and revolute joints, respectively. The
controller could asymptotically stabilize both position and synchronization errors
10 Synchronization and Control of Multiagent Systems

to zero and did not require the explicit use of the system dynamic model. Su et al.
(2006) integrated a saturated proportional-integral (PI) synchronous control and PD
feedback for high-precision control of parallel manipulators.
Synchronization technology also exhibits unique advantages for multirobot for-
mation controls. Studies on formation controls of swarms of mobile robots have
received increasing attention in recent years. Examples of formation control tasks
include assignment of feasible formations, movement into a formation, maintenance
of formation shape, and switching between formations. Sun and Wang (2007) pro-
posed to use a cross coupling–based synchronization control strategy to address the
problem of multirobot control in time-varying formations. The basic idea is that a
team of mobile robots tracks each individual desired trajectory while synchroniz-
ing motions among the robots to keep a relative kinematics relationship in order to
maintain the desired time-varying formation. More sufficient results were reported
in Sun et al. (2009) with experimental studies.
To summarize, over the past decades, the synchronization control has exhibited
advantages in many control applications of multiagent systems with great impact.
It is expected that the synchronization control will continue to play important roles
in multiagent systems in the future and will show its unique features in some new
areas such as infrastructure networks in system biology, where multiple biological
materials need to be handled simultaneously in complex environments at microscale
or nanoscale.

1.3 Outline of the Book

This book will systematically introduce synchronization control theory and applica-
tions in numerous multiagent systems. The outline of the book is listed below.
In Chapter 2, the concept of synchronization control strategy will be introduced.
The analysis will start from identification of the synchronization constraint and con-
trol goal, and then will define synchronization errors that will be used to measure
synchronicity of multiple agents in motions. Synchronization errors will be used as
the control variable in the control design, which distinguishes the synchronization
control from traditional nonsynchronized controllers.
In Chapter 3, synchronization of multiple motion axes, which is the most basic
problem in multiagent synchronization, will be investigated first. A PD-type cross-
coupling controller will be developed for position synchronization of multiaxis
motions in setpoint position controls. A setpoint tracking controller is further devel-
oped by adding feedforward control terms to the PD-type synchronization control-
ler. The method is easy to implement in practice because it is model free.
In Chapter 4, a new control approach to position tracking synchronization
of multiple motion axes will be developed, which overcomes limitations of the
PD-type synchronization control in Chapter 3. A model-based synchronization
approach will be proposed by using the cross-coupling concept, by which posi-
tion and synchronization errors can approach zero asymptotically in trajectory
tracking. To address the model uncertainty problem, an adaptive synchroniza-
tion control is further developed by incorporating cross-coupling technology into
adaptive control architecture. A case study using an adaptive coupling control to
Introduction 11

operate a CNC integrated machine will be introduced as an industrial applica-

tion example.
In Chapter 5, an adaptive synchronization control approach will be applied to
the coordination of multiple manipulators in a robot assembly work cell. The key
to the success of this new coordination method is to ensure that each manipulator
tracks its desired trajectory while synchronizing its motion with other manipulators’
motions to maintain the required kinematic relationship among robots. The control-
ler, designed by incorporating the cross-coupling technology into adaptive control
architecture, successfully guarantees asymptotic convergence to zero if both position
tracking and synchronization errors are found simultaneously. This will be verified in
both simulations and experiments on multirobot assembly tasks.
In Chapter 6, the synchronization approach will be applied to multiaxis CNC
machine tools to minimize contouring errors. Contouring errors are caused by all
motion axes in the machining process, and they significantly affect CNC machin-
ing accuracy. With the synchronization approach, contouring errors can be posed
as position synchronization errors, and a decentralized tracking controller will be
designed to stabilize position and contour errors simultaneously. The proposed con-
troller does not require significant use of the system dynamic models, which simpli-
fies actual implementation. Experiments performed on a real CNC machine tool
will demonstrate effectiveness of the synchronization control in reducing contouring
errors.
In Chapter 7, the synchronization approach will be applied to controls of robot
parallel manipulators for performance improvement. The relative complex mecha­
nism makes the precise control of parallel manipulators more difficult. The
cross-coupled synchronization control approach will provide a unique solution to
coordinate motions of all the actuators of parallel manipulators and thus improve
the position accuracy. A decentralized trajectory tracking controller will be intro-
duced, which feeds back both position and synchronization errors, formed with
a combination of feedforward, feedback, and saturation control. Experimental
results demonstrate improved performance with the proposed synchronous control
design.
Finally, in Chapter 8, the synchronization approach will be used to solve a multi-
robot formation control problem. The formation control problem will be posed as a
synchronization control problem, and based on this, the synchronization control goal
and formation errors can be defined. A synchronous controller will be designed to
control each robot’s translation to guarantee that both position and synchronization
errors approach zero asymptotically. Simulations and experiments are performed to
demonstrate the effectiveness of the proposed synchronization control approach in
the formation control tasks.

1.4 Summary
The intent of this chapter was to familiarize the reader with the history and current
application trends of multiagent systems and their coordinative controls. Numerous
challenging problems encountered by various coordination tasks are listed. It was
recognized that the synchronization control could provide unique advantages and
12 Synchronization and Control of Multiagent Systems

opportunities to solve various multiagent coordination problems. The cross cou-

pling–based synchronization control technologies have been reviewed, especially in
the fields of robotics and manufacturing automation.
The following is a list of major concepts, facts, and skills associated with this
chapter:

• Multiagent system and its coordination in both civil and military applications
• Main challenging problems encountered by multiagent coordination
• Synchronization concept and unique advantages for solving multiagent
coordination problems
• Cross-coupling strategy used in motion synchronization
• Reviews of synchronization studies in multiaxis motion controls, CNC
machining, coordination of multiple manipulators, high-precision controls of
parallel manipulators, and formation controls of swarms of mobile robots

References
Arimoto, S., Miyazaki, F., and Kawamura, S. 1987. Cooperative motion control of multiple
robot arms or fingers. IEEE International Conference on Robotics and Automation,
Raleigh, 1407–1412.
Balch, T., and Arkin, R. 1998. Behavior-based formation control for multi-robot systems.
IEEE Transactions on Robotics and Automation 14(6):926–939.
Beard, R. W., Lawton, H., and Hadaegh, F. Y. 2001. A coordination architecture for spacecraft
formation control. IEEE Transactions on Control System Technology 9(6):777–790.
Belta, C., and Kumar, V. 2004. Optimal motion generation for groups of robots: A geometric
approach. ASME Journal of Mechanical Design 126(1):63–70.
Berman, S., Edan, Y., and Hamshidi, M. 2003. Navigation of decentralized autonomous
automatic guided vehicles in material handling. IEEE Transactions on Robotics and
Automation 19(4):743–749.
Borenstein, J. 1995. Control and kinematic design of multi-degree-of-freedom mobile robots
with compliant linkage. IEEE Transactions on Robotics and Automation 11(1):21–35.
Borenstein, J., and Koren, Y. 1987. Motion control analysis of a mobile robot. ASME Journal
of Dynamic Systems, Measurement, and Control 109(2):73–79.
Chen, J., Sun, D., Yang, J., and Chen, H. 2010. Leader-follower formation control of multiple
nonholonomic mobile robots incorporating a receding-horizon scheme. International
Journal of Robotics Research 29(6):727–747.
Chiu, T. C., and Tomizuka, M., 1992. Motion synchronization in multiple axes mechanical
systems. ASME Paper 92-WA/DSC-13.
Chiu, T. C., and Tomizuka, M. 1998. Coordinated position control of multi-axis mechanical sys-
tems. ASEM Journal of Dynamic Systems, Measurement, and Control 120(3):389–393.
Chiu, T. C., and Tomizuka, M. 2001. Contouring control of machine tool feed drive systems:
A task coordinate frame approach. IEEE Transactions on Control Systems Technology
9(1):130–139.
Chuang, H. Y., and Chang, Y.-C. 2000. A novel contour error compensator for 3–PRPS plat-
form, Journal of Robotic Systems 17(5):273–289.
Chung, S. J., and Slotine, J. J. E. 2009. Cooperative robot control and concurrent synchroniza-
tion of Lagrangian systems. IEEE Transactions on Robotics 25(3):686–700.
Das, A. K., Fierro, R., Kumar, V. J., Ostrowski, J. P., Spletzer, J., and Taylor, C. J. 2002.
A vision-based formation control framework. IEEE Transactions on Robotics and
Automation 18(5):813–825.
Introduction 13

Deasi, J. P., Kumar, V. J., and Ostrowski, P. 2001. Modeling and control of formations of nonholo-
nomic mobile robots. IEEE Transactions on Robotics and Automation 17(6):905–908.
Egerstedt, M., and Hu, X. 2001. Formation constrained multiagent control. IEEE Transactions
on Robotics and Automation 17(6):947–951.
El-Ali, Sorger, J. P. K., and Jensen, K. F. 2006. Cells on chips. Nature 442(7101):403–411.
Fax, J. A., and Murray, R. M. 2004. Information flow and cooperative control of vehicle for-
mations. IEEE Transactions on Automatic Control 49(9):1465–1476.
Feng, L., Koren, Y., and Borenstein, J. 1993. Cross–coupling motion controller for mobile
robots, IEEE Control System Magazine 13(6):35–43.
Fox, D., Burgard, W., Kruppa, H., and Thrun, S. 2000. A probabilistic approach to collabora-
tive multi-robot localization. Autonomous Robots 8(3):325–344.
Giulietti, F., Pollini, L., and Innocenti, M. 2000. Autonomous formation flight. IEEE Control
System Magazine 20(1):34–44.
Hsu, P. 1989. Control of multi-manipulator systems: trajectory tracking, load distribution,
internal force control and decentralized architecture. IEEE International Conference on
Robotics and Automation, Scottsdale, 1234–1239.
Huang, J., Farritor, S., Qadi, M. A., and Goddard, S. 2006. Localization and follow-the-
leader control of a heterogeneous group of mobile robots. IEEE/ASME Transactions on
Mechatronics 11(2):205–215.
Jadbabaie, A., Lin, J., and Morse, A. S. 2003. Coordination of groups of mobile autonomous
agents using nearest neighbor rules. IEEE Transactions on Automatic Control, 48(9):
988–1001.
Kamano, T., Suzuki, T., Iuchi, N., and Tomizuka, M. 1993. Adaptive feedforward controller for
synchronization of two axes positioning system. Transactions of Society of Instrument
and Control Engineers (SICE), Japan, 29(7):785–791.
Kang, W., Xi, N., and Sparks, A. 2000. Formation control of autonomous agents in 3D work-
space. IEEE International Conference on Robotics and Automation, San Francisco, CA,
1755–1760.
Koivo, A. J., and Unseren, M. A. 1991. Reduced order model and decoupled control architec-
ture for two manipulators holding a rigid object. ASME Journal of Dynamic Systems,
Measurement, and Control 113(4):646–654.
Koren, Y. 1980. Cross-coupled biaxial computer controls for manufacturing systems. ASME
Journal of Dynamic Systems, Measurement, and Control 102(4):265–272.
Koren, Y., and Lo, C. C. 1991. Variable-gain cross-coupling controller for contouring. Annals
of CIRP 40(1):371–374.
Koren, Y., and Lo, C. C. 1992a. Evaluation of servo-controllers for machine tools. Proceedings
of American Control Conference, Chicago, Illinois, 370–374.
Koren, Y., and Lo, C. C. 1992b. Advanced controllers for feed drives. Annals of CIRP,
689–698.
Kosuge, K. et al. 1993. Decentralized coordinated motion control of multirobots. RSJ Annual
Conference, 705–706.
Kulkarni, P., and Srinivasan, K. 1985. Cross-coupled compensators for contouring control of
multi-axial machine tools, in Proceedings of the 13th North American Manufacturing
Research Conference, California, 558–566.
Kulkarni, P., and Srinivasan, K. 1989. Optimal contouring control of multi-axial feed drive
servomechanisms. ASME Journal of Engineering for Industry 111:140–148.
Kulkarni, P. K., and Srinivasan, K. 1990. Cross-coupled control of biaxial feed drive ser-
vomechanisms. ASME Journal of Dynamic Systems, Measurement, and Control
112(2):225–232.
Lawton, J. R., Beard, R. W., and Young, B. J. 2003. A decentralized approach to formation
maneuvers. IEEE Transactions on Robotics and Automation 19(6):933–941.
14 Synchronization and Control of Multiagent Systems

Lee, H. C., and Jeon, G. J. 1998. A neuro-controller for synchronization of two motion axes.
International Journal of Intelligent Systems 13(6):571–586.
Lewis, M. A., and Tan, K. H. 1997. High precision formation control of mobile robots using
virtual structures. Autonomous Robot 4:387–403.
Liu, Y. H., Arimoto, S., and Ogasawara, T. 1996. Decentralized cooperation control: non-
communication object handling. IEEE International Conference on Robotics and
Automation, Minneapolis, 2414–2419.
Luh, J. Y. S., and Zheng, Y. F. 1987. Constrained relations between two coordinated industrial
robots for motion control. International Journal of Robotics Research 6(3):60–70.
McInnes, C. R. 1995. Autonomous ring formation for a planar constellation of satellites. AIAA
Journal of Guidance, Control and Dynamics 18(5):1215–1217.
McNab, R., and Tsao, T. 1994, November. Multiaxis contour tracking: a receding time horizon
linear quadratic optimal control approach. Dynamic System and Control 2.
McNab, R. J., and Tsao, T. C. 2000. Receding horizon linear quadratic optimal control for
multi-axis contour tracking. ASME Journal of Dynamic Systems, Measurement, and
Control 122(2):375–381.
Mesbahi, M., and Hadaegh, F. 2001. Formation flying of multiple spacecraft via graphs,
matrix inequalities, and switching. AIAA Journal of Guidance, Control, and Dynamics
24:369–377.
Moreau, L. 2005. Stability of multiagent systems with time–deper client communication links.
IEEE Transactions on Automatic Control, 50(2):169–182.
Moore, P. R., and Chen, C. M. 1995. Fuzzy logic coupling and synchronized control of mul-
tiple independent servo-drives, Control Engineering Practice 3(12):1697–1708.
Naumovic, M. 1999. Cross-coupled motion controller for two cooperating robot arms.
Proceedings of IEEE International Symposium on Industrial Electronics, Bled, Slovenia,
2:909–913.
Ogren, P., Fiorelli, E., and Leonard, N. E. 2004. Cooperative control of mobile sensor net-
works: adaptive gradient climbing in a distributed environment. IEEE Transactions on
Automatic Control 40(8):1292–1302.
Olfati-Saber, R. 2006. Flocking for multiagent dynamic systems: algorithms and theory. IEEE
Transactions on Automatic Control 51(3):401–420.
Olfati-Saber, R., and Murray, R. M. 2004. Consensus problems in networks of agents with switch-
ing topology and time-delays. IEEE Transactions on Automatic Control 49(9):101–115.
Parker, L. E. 1998. ALLIANCE: an architecture for fault tolerant multirobot cooperation.
IEEE Transactions on Robotics and Automation 14(2):220–240.
Ren, W., and Beard, R. W. 2005. Consensus seeking in multiagent systems under dynami-
cally changing interaction topologies. IEEE Transactions on Automatic Control
50(5):655–661.
Rodriguez-Angeles, A., and Nijmeijer, H. 2004. Mutual synchronization of robots via esti-
mated state feedback: a cooperative approach. IEEE Transactions on Control Systems
Technology 12(4):542–554.
Sepulchre, R., Paley, D., and Leonard, N. E. 2007. Stabilization of planar collective motion:
all-to-all communication. IEEE Transactions on Automatic Control 52(5):811–824.
Sowinski, S., Jolly, C., Berninghausen, O., Purbhoo, M. A., Chauveau, A., Köhler, K., Oddos,
S., Eissmann, P., Brodsky, F. M., Hopkins, C., Önfelt, B., Sattentau, Q., and Davis, D. M.
2008. Membrane nanotubes physically connect T cells over long distances presenting a
novel route for HIV-1 transmission. Nature Cell Biology 10(2):211–219.
Srinivasan, K., and Fosdick, R. 1988. Multivariable analysis and controller design for coor-
dinated multi-axial motion control, in Proceedings of American Control Conference,
Atlanta, GA, 95–101.
Stilwell, D. J., and Bishop, B. E. 2000. Platoons of underwater vehicles. IEEE Control System
Magazine 20(1):45–52.
Introduction 15

Su, Y. X., Sun, D., Ren, L., and Mills, J. K. 2006. Integration of saturated PI synchronous
control and PD feedback for control of parallel manipulators. IEEE Transactions on
Robotics 22(1):202–207.
Sun, D. 2003a. Position synchronization of multiple motion axes with adaptive coupling con-
trol. Automatica 39(6):997–1005.
Sun, D. 2003b. Adaptive coupling control of two working operations in CNC inte-
grated machines. ASME Journal of Dynamic Systems, Measurement, and Control
125(4):662–665.
Sun, D., Feng, G., Lam, C. M., and Dong, H. N. 2005. Orientation control of a differen-
tial mobile robot through wheels’ synchronization. IEEE/ASME Transactions on
Mechatronics 10(3):345–351.
Sun, D., Lu, R., Mills, J. K., and Wang, C. 2006. Synchronous tracking control of paral-
lel manipulators using cross-coupling approach. International Journal of Robotics
Research 25(11):1137–1147.
Sun, D., and Mills, J. K. 2002. Adaptive synchronized control for coordination of multirobot
assembly tasks. IEEE Transactions on Robotics and Automation 18(4):498–510.
Sun, D., Shao, X., and Feng, G. 2007. A model-free cross-coupled control for position syn-
chronization of multi-axis motions: theory and experiments. IEEE Transactions on
Control Systems Technology 15(2):306–314.
Sun, D., and Tong, M. C. 2009. A synchronization approach for the minimization of con-
touring errors of CNC machine tools. IEEE Transactions on Automation Science and
Engineering 6(4):720–729.
Sun, D., and Wang, C. 2007. Controlling swarms of mobile robots for switching between
formations using synchronization concept. IEEE International Conference on Robotics
and Automation, Roma, Italy, 2300–2305.
Sun. D., Wang, C., Shang, W., and Feng, G. 2009. A synchronization approach to trajectory
tracking of multiple mobile robots while maintaining time-varying formations. IEEE
Transactions on Robotics 25(5):1074–1086.
Tanner, H. G., Loizou, S. G., and Kyriakopoulos, K. J. 2003. Nonholonomic navigation and
control of multiple mobile manipulators. IEEE Transactions on Robotics and Automation
19(1):53–64.
Tarn, T. J., Bejczy, A. K., and Yun, X. 1986. Coordinated control of two arms. IEEE International
Conference on Robotics and Automation, San Francisco, 1193–1202.
Tomizuka, M., Hu, J. S., and Chiu, T. C. 1992. Synchronization of two motion control axes
under adaptive feedforward control. ASME Journal of Dynamic Systems, Measurement,
and Control 114(6):196–203.
Wen, J. T., and Delgado, K. K. 1992. Motion and force control of multiple robotic manipula-
tors. Automatica 28(4):729–743.
Xi, N., Tarn, T. J., and Bejczy, A. K. 1996. Intelligent planning and control for multirobot
coordination: an event-based approach. IEEE Transactions on Robotics and Automation
12(3):439–452.
Yang, L. F., and Chang, W. H. 1996. Synchronization of twin-gyro precession under cross-
coupled adaptive feedforward control. Journal of Guidance, Control and Dynamics
19(3):534–539.
Yeh, S. S., and Hsu, P. L. 2003. Analysis and design of integrated control for multi-axis motion
systems. IEEE Transactions on Control Systems Technology 11(3):375–382.
Yoshikawa, T., Sugie, T., and Tanaka, M. 1988. Dynamic hybrid position/force control of
robot manipulator-controller design and experiment. IEEE Transactions on Robotics
and Automation 4(6):699–705.
Yun, X., Sarkar, N., and Kumar, V. J. 1997. Dynamic control of 3-D rolling contacts in two-
arm manipulation. IEEE Transactions on Robotics and Automation 13(3):364–376.
 2 Synchronization Strategy

Abstract
Synchronization is an important methodology existing in human life every-
where. This chapter will introduce the concept of synchronization and some gen-
eral applications of synchronization in different aspects. The key of this chapter
is application of the synchronization concept to the coordination of multiagent
motions. This analysis will start from discussions on how to identify synchro-
nization constraints and goals, and then go to definitions of synchronization
errors. Synchronization errors will be used to measure the synchronicity of mul-
tiple agents in the coordinated motions, and will be used as the control variables
in the synchronization controller to be developed in the next chapters.
After reading this chapter, readers will understand the idea of synchronization
strategy for coordination of multiagent systems, which will lay a solid founda-
tion for the understanding of synchronous controls for various applications in
the following chapters.

2.1 Concept of Synchronization
Synchronization is timekeeping that requires the coordination of events to operate a
system in unison. The purposes of synchronization can be summarized as follows:

• Safety: Many multiagent systems require coordination among multiple

agents to avoid conflict with each other, which has become a necessary con-
dition to guarantee safety. Figure 2.1 illustrates an example that trains are
in synchronization on railway timetables to avoid any possible conflict—
synchronization among trains plays a key role in guaranteeing safety.
• Precision: When multiple agents work together for a common goal and can
only deliver a precise result when they operate in a coordinative way, syn-
chronization will play a key role in guaranteeing precision of the process.
A typical example to show precision due to synchronization is a clock. On
a clock, the hour hand, minute hand, and second hand are synchronized, as
seen in Figure 2.2.
• Efficiency: In a similar way to precision, synchronization helps ensure effi-
ciency of many processes. An example is that industrial gears are synchro-
nized, as shown in Figure 2.3, where poor synchronization among gears
will reduce efficiency or even lead to failure of the task.
• Entertainment: Synchronization has been used for coordination between
image and sound in entertainment. Figure 2.4 illustrates the example

17
18 Synchronization and Control of Multiagent Systems

Figure 2.1 Trains in synchronization on railway timetables.

that image and sound are synchronized on the film. Poor synchroniza-
tion will lead to mismatch between image and sound and thus affect the
show performance.
• Communication: Synchronization has also been found as an important tool
in communication. Figure 2.5 shows that the devices on a network are syn-
chronized to process information.

Figure 2.2 The hour hand, minute hand, and second hand of clocks are in
synchronization.
Synchronization Strategy 19

Figure 2.3 Industrial gears in synchronization.

This book will emphasize motion synchronization of multiple agents. In a multi-

agent coordination system, motion synchronization will ensure states of all agents to
correlatively follow a common regulation while each agent has an individual motion
track; hence, motion synchronization acts as the timekeeper to operate the whole
system in unison. The importance of motion synchronization has been found in
many applications, such as contouring control of multiple computer numerical con-
trol (CNC) machining axes, coordination of multiple robot manipulators in assembly
tasks, and formation control of networked robots, to name a few. In multiaxis CNC

Figure 2.4 Image and sound are synchronized on the film.

20 Synchronization and Control of Multiagent Systems

Figure 2.5 Devices on a network are synchronized.

machining, for example, a synchronization controller can be used to control each

machining axis to follow its desired trajectory by converging the position/tracking
error to zero, and meanwhile, to synchronize the motion of each axis with that of
others to minimize the differential position errors among axes, which can subse-
quently reduce contouring errors (Sun and Tong 2009).
In the following, a motion synchronization strategy will be introduced, which
includes identification of synchronization constraint, definition of synchronization
control goal, and definition of synchronization error. Both synchronization control
goal and synchronization error will be used in the design of the synchronization
controller, as described in the next chapters.

2.2 Synchronization Control Goal

Consider that a multiagent system with n agents is engaged in an event that requires
motion coordination of all agents. Synchronization control aims to synchronize
motions of all agents such that the agents maintain a certain kinematics relationship, as
required by the coordination task. Regulating multiple agents to maintain the required
kinematics relationship can be presented as guiding and positioning the agents along
the boundary (or curve) of a multidimensional compact set. Denote S (ρ, t ) as a time-
varying desired shape in such a compact set, where ρ denotes a state vector and t the
time. The boundary of S (ρ, t ) is parameterized by a curve, denoted by ∂S (ρ, t ) = 0.
Denote xi (t ) and xid (t ) as the generalized state coordinate and its desired value
of the ith agent, respectively, where i = 1, … , n , representing an index. Note that xid (t )
must be defined to locate on the desired curve such that ∂S ( xid , t ) = 0. The state error
of the ith agent is represented as

ei (t ) = xid (t ) − xi (t ) (2.1)
Synchronization Strategy 21

The control objective is converging each agent to its desired state xid (t ) such that
ei → 0 as time t → ∞, and meanwhile, maintaining the agent on the desired curve
such that ∂S ( xi , t ) = 0. The former is called the first task, which represents a tradi-
tional motion control task aimed to achieve the motion control goal of each indi-
vidual agent. The later is called the second task, which represents a synchronization
control effort aiming for coordinating multiagent motions to satisfy the required
kinematic relationship among the agents. Both Task 1 and Task 2 are required to be
accomplished simultaneously, with one motion controller.
To fulfill the second task requirement as mentioned above, all the agents must
be subject to a common constraint that has been mathematically represented as
∂S (ρ, t ) = 0. Although such a constraint may appear in different formulations for
different synchronization tasks, people try to find a common character associated
with this constraint, which could link all the coordination requirements together in
a common expression.
Without loss of generality, one can express the constraint that the coordinate xi (t )
of the ith agent must meet in the following form:

∂S ( xi , t ) = 0 :{xi (t ) = Ai (t )C (t ) + Bi (t )} (2.2)

where Ai (t ) is a constraint matrix formulated based on the desired boundary, which

depends on the characteristics of the ith agent; C (t ) is a common vector that is
applicable to all the agents and does not on individual agents; and Bi (t ) is an offset
of the ith agent.
It is seen from (2.2) that all the agents can be linked via the common vector C (t ),
which requires a linear mapping from xi (t ) to the common vector C (t ), denoted by
{xi (t )}  C (t ), to be realized. To ensure a unique linear mapping {xi (t )}  C (t ) , the
inverse of the constraint matrix Ai (t ) must exist. Because the constraint matrix Ai (t )
is formulated based on the desired boundary ∂S ( xi , t ) = 0, it is largely determined by
the topology design for the given formation task. For simplicity, this book assumes
that the desired boundary ∂S ( xi , t ) = 0 for each agent can be properly designed such
that the inverse of Ai (t ) exists.
The following assumption is thus given:

Assumption 2.1: The desired boundary ∂S ( xi , t ) = 0 is properly designed such that

the inverse of the constraint matrix Ai (t ) exists.
Based on Equation (2.2) and Assumption 2.1, it follows that

Ai−1 (t )( xi (t ) − Bi (t )) = C (t ) (2.3)
Equation (2.3) gives a mathematical relationship between coordinates xi (t ) and
the common vector C (t ). If considering all agents with indices i = 1, … , n , the fol-
lowing constraint equations can be established:

A1−1 (t )( x1 (t ) − B1 (t )) = A2−1 (t )( x2 (t ) − B2 (t )) =  = An−1 (t )( x n (t ) − Bn (t )) = C (t ) (2.4)

Equation (2.4) implies that a mutual connection among all agents can be estab-
lished via the common vector C (t ). Because Ai (t ) is determined by the boundary
22 Synchronization and Control of Multiagent Systems

curve ∂S ( xi , t ) = 0, Equation (2.4) also provides a mathematical model for the bound-
ary curve on which all agents (i = 1, … , n ) locate.
The constraint Equation (2.4) must also be valid to the desired coordinate xid (t ),
namely,
A1−1 (t ) ( x1d (t ) − B1 (t ) ) = A2−1 (t ) ( x 2d (t ) − B2 (t ) ) =  = An−1 (t ) ( x nd (t ) − Bn (t ) ) = C (t ) (2.5)

Subtracting (2.4) from (2.5) and utilizing Equation (2.1) yields

A1−1 (t )e1 (t ) = A2−1 (t )e2 (t ) =  = An−1 (t )en (t ) (2.6)

Furthermore, define ci (t ) ≡ Ai−1 (t ) as the coupling parameter of the ith agent.

Then, rewrite Equation (2.6) as
c1 (t )e1 (t ) = c2 (t )e2 (t ) =  = cn (t )en (t ) (2.7)

Equation (2.7) represents a relationship that all the position errors must be regu-
lated to satisfy in order to ensure that multiple agents meet the coordination require-
ment. It is therefore used as the synchronization control goal.
In the following, a few examples will be given to show how the synchronization
control goal can be derived from actual case analysis.

Example 2.1

Consider two manipulators holding a rigid beam in a trajectory tracking task

(Sun and Mills 2002), as shown in Figure 2.6. It is a requirement that the distance
between positions of the two end-effectors of manipulators must remain the same
so as not to deform the beam, thus avoiding damage to the beam or manipulators.
The position coordinates of the two end-effectors, denoted by x1(t ) and x2(t ),
respectively, are subject to the following constraints:
1
x1(t ) = xo (t ) + d
2
1
x2(t ) = xo (t ) − d
2

x1(t) x0(t) x2(t)

d
Beam

Manipulator 1 Manipulator 2

Figure 2.6 Two robot manipulators grasping a beam.

Synchronization Strategy 23

where xo (t ) denotes the position coordinate of the geometrical center of the

grasped beam, and d denotes the distance between the two grasping positions
on the beam.
By referring to the constraint formulation as given in Equation (2.2), it follows that

A1(t ) = A2(t ) = I

C (t ) = xo (t )

1
B1(t ) = d
2
1
B2(t ) = − d
2

where I is a unit matrix. Define c1(t ) = c 2(t ) ≡ I . The following synchronization

control goal is then obtained as

e1(t ) = e2(t )
■

Example 2.2

Consider that a differential mobile robot with two driving wheels tracks a curved
path (Sun and Mills 2002), as shown in Figure 2.7. Denote the radii of the desired
curves followed by the two driving wheels as r1(t ) and r2(t ), respectively. The dis-
placements of the two driving wheels (left and right ones), denoted by x1(t ) and
x2(t ), are subject to the following constraints:

x1(t ) x2(t )
= = β(t )
r1(t ) r2(t )

x1(t)

x2(t)

r2(t)

Mobile robot
r1(t)

β(t)

Figure 2.7 A differential mobile robot in a curved path.

24 Synchronization and Control of Multiagent Systems

where β(t ) denotes the span angle. By referring to Equation (2.2), one obtains

A1(t ) = r1(t )

A2(t ) = r2(t )

C(t ) = β(t )

B1(t ) = B2(t ) = 0

Define

c1(t ) ≡ A1−1(t ) = r1−1(t )

c 2(t ) ≡ A2−1(t ) = r2−1(t )

The synchronization control goal is then obtained as follows:

c1(t )e1(t ) = c 2(t )e2(t )

■

Example 2.3

Consider n moving agents that are required to maintain a time-varying ellipse

formation (Sun et al. 2009), as shown in Figure 2.8. The position coordinate of

100
Target shape
80

60
Initial shape
40

20
Y (m)

0
Initial position
–20 of robot i

–40

–60

–80
Final position of robot i
–100
–60 –40 –20 0 20 40 60
X (m)

Figure 2.8 Multiple agents moving in ellipse formation.

Synchronization Strategy 25

each agent, denoted by xi (t ), is required to meet the following constraint:

 cos ϕ i (t )   a(t ) 
xi (t ) =   
 sin ϕ i (t )   b(t ) 


where a(t ) and b(t ) denote the longest and the shortest radii of the ellipse, respec-
)sin α i
tively; ϕ i (t ) = tanh( ab((tt)cos α i ) ; and α i denotes the angle of the ith robot lying on the
ellipse with respect to the center of the ellipse.
By referring to Equation (2.2), one has the following results:

 cos ϕ i (t ) 
Ai (t ) =  
 sin ϕ i (t ) 


 a(t ) 
C (t ) =  
 b(t ) 
Bi (t ) = 0

Assume that the robots are not located in the longest or the shortest axes of
the ellipse such that the inverse of Ai exists. Define ci (t ) ≡ Ai−1(t ) . One then could
establish the synchronization control goal as follows:

c1(t )e1(t ) = c 2(t )e2(t ) =  = cn (t )en (t )

■

The above examples show how to establish a synchronization control goal based
on the synchronization constraint as developed in Equation (2.2). If the required
coordination can be mathematically modeled with a boundary curve ∂S (ρ, t ) = 0 sub-
ject to the constraint in Equation (2.2), a synchronization control goal can then be
established, based on which a synchronization controller can be further developed to
achieve the coordination goal.

2.3 Synchronization Errors
After the synchronization control goal is established, a synchronization controller
will then be developed to achieve this goal. A question that is naturally generated
prior to the controller development is how to measure synchronicity of multiple
agents in synchronization. There is a need to find criteria that can be used to evalu-
ate the synchronization performance.
The concept of synchronization error is hereby introduced. In literature of
cross-coupling control, the differential state errors among agents were controlled.
In Tomizuka et al. (1992), for example, the differential velocity error between two
motion axes, namely x1 − x2 , was driven to zero, where xi denotes the velocity of the
ith motion axis. In Feng et al. (1993), the differential displacement error between two
driving wheels of a mobile robot, defined as c L q L − c R q R , was minimized to zero,
26 Synchronization and Control of Multiagent Systems

where q L and q R denote the displacements of the left and the right driving wheel,
respectively, and c L and c R are correction factors to overcome the curve displace-
ment difference. These examples set a reference that motions of multiple agents can
be synchronized through cross-coupling control of state errors among the agents.
Synchronization error may be defined in a similar way to the differential state
error in cross-coupling controls. Also, it must be suited to multiple agents (not
only two agents). The idea of defining synchronization error can be summarized
as follows:

1. Synchronization error should be defined based on the synchronization con-

trol goal in Equation (2.7).
2. The concept of the differential state error in cross-coupling controls can be
utilized in the definition of synchronization error.
3. The definition of synchronization error should not be limited to the scal-
ability of agents (i.e., it is suited to multiple agents, not just two agents).
4. The definition of synchronization error should include states of agents as
well as factors that produce constraints to states in synchronization (e.g.,
the correction factors c L and c R in cross-coupling control by Feng et al.,
1993).

Synchronization error will be defined to fulfill the above conditions. When the syn-
chronization goal (2.7) is well achieved, the following equations hold automatically:

c1 (t )e1 (t ) − c2 (t )e2 (t ) = 0

c2 (t )e2 (t ) − c3 (t )e3 (t ) = 0
(2.8)


cn (t )en (t ) − c1 (t )e1 (t ) = 0

Equation (2.8) implies that for any agent pair with indices i and i + 1, its differential
state error with the cross-coupling parameters, expressed by ci (t )ei (t ) − ci +1 (t )ei +1 (t ) ,
is zero when the synchronization goal is achieved. In other words, such a differential
state error can be used to measure synchronicity of a multiagent in synchronization:
if it is zero, the synchronization is perfect; otherwise, synchronization error exists.
Therefore, synchronization error may be defined as a subset of all possible pairs
of two neighboring agents in the following way:

ε1 (t ) = c1 (t )e1 (t ) − c2 (t )e2 (t )

ε 2 (t ) = c2 (t )e2 (t ) − c3 (t )e3 (t )
(2.9)

ε n (t ) = cn (t )en (t ) − c1 (t )e1 (t )
Synchronization Strategy 27

Line formation

Agents 1 2 …… n

Figure 2.9 Multiple agents in a line formation.

where εi (t ) denotes the synchronization error of the ith agent, and the cross-coupling
parameter ci (t ) helps construct the relationship between two agents in synchroniza-
tion. Obviously, if the synchronization error εi = 0 for all i = 1,..., n , the synchroni-
zation goal (2.7) is achieved automatically.
It is worth noting that the synchronization error εi (t ) as defined above is used to
measure the synchronization effect, which is not equivalent to the state error ei (t ).
For systems having closed-loop chain structures, employment of synchronization
error provides each agent with motion information both from itself and from the oth-
ers; hence, motions of all the networked agents are coordinated.
Agents can be coded in terms of their physical positions on the team when per-
forming the coordination task. A principle that may be followed is that the two physi-
cally close agents are coded as two neighbors, for easy sensorial connection. As a
result, all the agents in the group are linked as a whole, either directly or indirectly.
Now the control problem becomes to drive both the state error ei (t ) and the syn-
chronization error εi (t ) to zero. The ith agent can be designed to approach its desired
state xid (t ) while synchronizing its motion with those of two neighboring agents
with indices i – 1 and i + 1. In this way, the control of each agent requires only the
information of its two neighbors but not all the agents in the group; hence, the imple-
mentation is simplified.
In some cases, the two neighborly coded agents may be spatially far and hence
not sensorially connected. This usually happens when the desired shape S (ρ, t ) is
not geometrically closed. Figure 2.9 illustrates an example that multiple agents
are required to maintain a line formation, where the first agent (agent 1) and the
last agent (agent n) are far between each other but coded as two neighbors. If the
sensorial problem between these two agents cannot be solved, maintaining direct
synchronization between these two agents will be very difficult and may not be
necessary. (Instead, one may simply set the synchronization error between these
two agents to zero.) Due to the fact that these two agents still synchronize with the
other agents, all agents in the group (including these two agents) are still linked as
a whole.

2.4 Summary
The synchronization strategy used for coordinating motions of multiagent systems
is presented in this chapter. The goal is to help readers with different backgrounds
reach a common level and help them become aware of the fundamental issues and
major analytical procedures in modeling a general coordination problem as a syn-
chronization control problem.
28 Synchronization and Control of Multiagent Systems

The following summarizes the major concepts, facts, and skills associated with
this chapter:

• The basic concept and general applications of the synchronization technol-

ogy in our lives
• Use of a time-varying desired shape and its boundary to model kinematics
constraint as required by multiagents in coordination
• Establishment of synchronization constraint and control goal
• Definition of synchronization error

References
Feng, L., Koren, Y., and Borenstein, J. 1993. Cross–coupling motion controller for mobile
robots. IEEE Control Systems Magazine 13(6):35–43.
Sun, D., and Mills, J. K. 2002. Adaptive synchronized control for coordination of multirobot
assembly tasks. IEEE Transactions on Robotics and Automation 18(4):498–510.
Sun, D., and Tong, M. C. 2009. A synchronization approach for the minimization of con-
touring errors of CNC machine tools. IEEE Transactions on Automation Science and
Engineering 6(4):720–729.
Sun, D., Wang, C., Shang, W., and Feng, G. 2009. A synchronization approach to trajectory
tracking of multiple mobile robots while maintaining time-varying formations. IEEE
Transactions on Robotics 25(5):1074–1086.
Tomizuka, M., Hu, J. S., and Chiu, T. C. 1992. Synchronization of two motion control axes
under adaptive feedforward control. ASME Journal of Dynamic Systems Measurement
and Control 114(6):196–203.
 3 Model-Free
Synchronization Control
of Multiple Motion Axes
Abstract
In this chapter, a model-free synchronization approach to the coordination
of multiaxis motions will be introduced. Synchronization of multiple motion
axes represents a simple but fundamental issue in multiagent synchronization.
The analysis will start from a model-free cross-coupling setpoint control for
position synchronization of multiaxis motions in moving along a straight line.
A proportional derivative (PD)–type synchronized controller will be given.
Then, a setpoint tracking controller will be introduced by adding feedforward
control to the PD-type synchronized controller. Because the introduced meth-
ods are model free or do not heavily depend on the system models, they are
easily implemented in practice. Experimental examples are given to show the
effectiveness of the approach.

3.1 Problem Statement
With the ever-increasing demand for greater productivity and lower cost, there is
tremendous pressure to achieve rapid development with high performance in modern
manufacturing. Those manufacturing devices such as surface mounting technology
(SMT) machines or computer numerical control (CNC) machine tools are accord-
ingly required to have all machine axes move simultaneously or synchronously for
either reducing work-in-progress or complex part machining. Poor synchronization
of relevant motion control axes results in diminished dimensional accuracy of the
workpiece or in unusable products.
The cross-coupling technology, as introduced by Koren (1980) and Tomizuka
et al. (1992), provides advantages and opportunities to improve synchronization per-
formance. Over the past decades, the cross-coupling concept has been widely used in
multiaxis motions and other applications, such as reducing the contour error of CNC
machines (Rodriguez–Angeles and Nijmeijer 2004; Yan et al. 2005; Yeh and Hsu
2002; Zhong et al. 2002). The cross-coupling concept was incorporated into adaptive
control architecture to solve position synchronization of multiple axes (Sun 2003).
A model-free variable-gain cross-coupling controller was introduced for a general
class of contours (Koren and Lo 1991), but the effect of a time-varying gain in the
cross-coupling controller to system stability and the cross-coupling effect on overall

29
30 Synchronization and Control of Multiagent Systems

z
3D moving straight line

d(t)

Moving
platform

β
y

Figure 3.1 A platform moving along a straight line.

system dynamics are yet to be examined (Chiu and Tomizuka 2001). Many other
works can be found to reduce contour errors in CNC machining applications.
In many motion control applications, synchronization of multiaxis motions
plays a fundamental role in guaranteeing precision. Figure 3.1 illustrates an exam-
ple that a platform moves along a straight line in a framework with x-, y-, and
z-axes. Because the platform is driven by three actuators along the three axes, its
motion precision heavily depends on the synchronization of three-axis motions.
Let x1, x 2, and x3 be the position coordinates along x-, y-, and z-axes, respec-
tively. Denote ∂S ( xi , t ) = 0 as the boundary that the coordinate xi should locate
on, as introduced in Chapter 2. To guarantee the platform moves along the desired
straight line precisely, position coordinates of the three axes must meet the follow-
ing constraint relationship:

 ∂S ( x1 , t ) = 0 :{x1 = d (t ) cos β cos θ}


 ∂S ( x2 , t ) = 0 : {x2 = d (t ) cos β sin θ} (3.1)
 ∂S ( x , t ) = 0 :{x = d (t )sin β}
 3 3

where d (t ) denotes the moving distance along the straight line, and θ and β are
constant angles as shown in Figure 3.1, which are determined by the given straight
line. In the following, it will be shown that the accurate motion control problem can
be posed as a synchronization control problem.
Referring to the result in Chapter 2, the constraints in (3.1) can be presented in
the following form:

∂S ( xi , t ) = 0 :{xi (t ) = Ai (t )C (t ) + Bi (t )} (3.2)
Model-Free Synchronization Control of Multiple Motion Axes 31

with the following definitions

A1 = cos β cos θ

A2 = cos β sin θ

A3 = sin β

B1 (t ) = B2 (t ) = B3 (t ) = 0

C (t ) = d (t )

To ensure that the linear mapping {xi (t )}  C (t ) exists uniquely, the inverse of the
constraint matrix Ai must exist. This condition can be fulfilled by properly giving the
desired task with angles θ and β. Then, the constraint (3.2) can be rewritten as

Ai−1xi (t ) = d (t ) (3.3)

Using (3.3), the synchronization constraint of xi (t ) in Figure 3.1 can be math-

ematically established as follows:

A1−1x1 (t ) = A2−1x 2 (t ) = A3−1x3 (t ) = d (t ) (3.4)

This constraint is also valid to the desired coordinate xid (t ), namely,

A1−1x1d (t ) = A2−1x 2d (t ) = A3−1x3d (t ) = d (t ) (3.5)

Define the position error of the ith axis as

ei (t ) = xid (t ) − xi (t ) (3.6)

Subtracting (3.4) from (3.5) and utilizing (3.6) yields

A1−1e1 (t ) = A2−1e2 (t ) = A3−1e3 (t ) (3.7)

Define ci ≡ Ai−1 as the coupling parameter of the ith axis. Because both θ and β
are constant, Ai is constant, and so is ci . Rewrite Equation (3.7) as

c1e1 (t ) = c2e2 (t ) = c3e3 (t ) (3.8)

Equation (3.8) represents the synchronization control goal in the task of Figure 3.1.
In order to guarantee the platform to track the desired straight line exactly, the syn-
chronization goal (3.8) must be achieved.
32 Synchronization and Control of Multiagent Systems

Note that if the platform moves along a curve rather than a straight line, the
angles θ and β in Figure 3.1 are not constant. Subsequently, the coupling parameter
ci becomes time varying.
Further, expand the above example to a more general case with n motion axes,
where each axis has coupling parameter ci . Thus, a generalized synchronization
control goal can be represented as

c1e1 (t ) = c2e2 (t ) =  = cnen (t ) (3.9)

In case all ci are the same, the synchronization goal (3.9) can be simplified as
e1 (t ) = e2 (t ) =  = en (t ) .

3.2 Position Synchronization Errors

and Control Strategy
Based on Chapter 2, position synchronization error represents a differential position
error among multiple motion axes, which can be used for measuring synchronicity.
Position synchronization error of a multiaxis can be determined based on the syn-
chronization control goal as given in Equation (3.9).
Define position synchronization error as a subset of all possible pairs of two
neighboring axes in the following way:

ε1 (t ) = c1e1 (t ) − c2e2 (t )

ε 2 (t ) = c2e2 (t ) − c3e3 (t )
(3.10)


ε n (t ) = cnen (t ) − c1e1 (t )

where εi (t ) denotes the position synchronization error of the ith axis, and ci
denotes the coupling parameter that reflects the constraint relationship between
two neighboring axes. Obviously, if the position synchronization error εi = 0 for
all i = 1, … , n , the synchronization control goal (3.9) is achieved automatically.
Rewrite (3.10) in the matrix format as

 ε1 (t )   c1 −c2 0  0   e1 (t ) 
    
 ε 2 (t )   0 c2 −c3  0   e2 (t ) 
  =         (3.11)
    
ε
 n−1 (t )   0  0 cn−1 −cn   n−1 (t )
e 
 ε n (t )   −c1 0  0 cn   en (t ) 
    
Model-Free Synchronization Control of Multiple Motion Axes 33

which can be further expressed as

ε(t ) = Te(t ) (3.12)

where ε(t ) = [ε1 (t ), ε 2 (t ), … , ε n (t )]T denotes a vector containing synchronization

errors of all axes, e(t ) = [e1 (t ), e2 (t ), … , en (t )]T denotes the position error vector, and
T denotes a transformation matrix. Note that when the moving trajectory is a straight
line as shown in Figure 3.1, T is constant.
The synchronization control objective is to design a synchronous controller that
can guarantee asymptotic convergence to zero of both the position error e(t ) and
the synchronization error ε(t ) . To achieve this goal, one needs to think about how
to link both position and synchronization errors into one control variable in the syn-
chronous controller.
Introduce the concept of coupled position error that links e(t ) and ε(t ) together
as expressed below:

E (t ) = e(t ) + αε(t ) (3.13)

where α is a control gain matrix that is diagonal and positive definite. Substituting
(3.12) into (3.13) yields

E (t ) = ( I + αT )e(t ) (3.14)

where I is a unit matrix.

Based on Equation (3.14), if the inverse of ( I + αT ) exists, E (t ) → 0 implies
e(t ) → 0. Further, e(t ) → 0 implies ε(t ) → 0 from (3.12). To guarantee that the
inverse of ( I + αT ) exists, the control gain α should be properly chosen. One easy
solution is to choose α to be small enough such that ( I + αT ) is positive definite and
has full rank.
In the following sections, detailed control algorithms will be given to drive the
coupled position error E (t ) to zero as time t → 0 .

3.3 Multiaxis Synchronization
in Setpoint Position Control
First, a relatively simple case of setpoint position control of moving a platform in a
straight line by driving multiple motion axes synchronously will be considered. In
the setpoint control, there is no need to consider the desired velocity during move-
ment, and thus, e = − x , where x = [ x1 x 2  x n ]T and the desired velocity x d
is set to zero. Because the angles θ and β remain the same when moving along a
straight line, the transformation matrix T is constant (see Section 3.2). Then, the
following equation can be derived from Equation (3.12):

ε = Te = −Tx (3.15)

34 Synchronization and Control of Multiagent Systems

Design a PD-type synchronization control law in the following format (Sun et al.
2007):

τ = K P E + K D E + ( I + αT )−1 K ee (3.16)

where K P , K D , and K e are all positive control gains. Obviously, the controller (3.16)
is model free. The third term of the right-hand side of (3.16) is introduced for system
stability, which will be detailed in the following analysis.
Consider the dynamics of a general mechanical system with n motion axes as
follows:

H ( x ) x + C ( x,x ) x = τ (3.17)

where H ( x ) = diag{Hi ( xi )} is the inertia of the multiaxis system, which is positive

definite; C ( x , x ) = diag{Ci ( xi , xi )} denotes the Coriolis and centripetal forces, and
H ( x ) − 2C ( x,x ) is skew-symmetric; and τ = [τ1 τ 2  τ n ] denotes the vector
of input torque. Note that both H ( x ) and C ( x,x ) satisfy the Lipschitz condition
(Filippov 1988).
Substituting (3.16) into the dynamics (3.17) yields the following closed-loop
dynamics:

H ( x ) x + C ( x,x ) x = K P E + K D E + ( I + αT )−1 K ee (3.18)

The following theorem will show the stability of the system under the control-
ler (3.16).

Theorem 3.1
The controller (3.16) guarantees e → 0 and ε → 0 as time t → ∞ , under the
following conditions:

1. The control gain α is properly chosen such that ( I + αT ) is positive and its
inverse exists.
2. The gain matrix K e is chosen to satisfy λ min {K e} = C1  x  +C2  x  ≥ c1 ×
 x  + c2  x  , where λ min {K e} is the minimum eigenvalue of the gain matrix
K e , C1 and C2 are positive constants, and c1 and c2 are Lipschitz constants.

Proof
Define a positive function as

1 T 1
V= e ( I + αT ) H ( x )e + E T K P E (3.19)
2 2
Model-Free Synchronization Control of Multiple Motion Axes 35

If the control gain α is chosen to be small enough such that eT ( I + αT ) H ( x )e ≥ 0,
the function (3.19) is positive-definite and can be used as a Lyapunov function can-
didate. Differentiating V with respect to time yields

1
V = eT ( I + αT ) H ( x )e + eT ( I + αT ) H ( x )e + E T K P E (3.20)
2

Multiplying both sides of (3.18) by E T yields

E T H ( x ) x + E T C ( x , x ) x = E T K P E + E T K D E + E T ( I + αT )−1 K ee (3.21)

Utilizing (3.12) through (3.14), Equation (3.21) can be rewritten as follows:

−eT ( I + αT ) H ( x )e − eT ( I + αT )C ( x , x )e = E T K P E + E T K D E + eT K ee (3.22)

Substituting (3.22) into (3.20) yields

1
V = eT ( I + αT )  H ( x ) − C ( x , x ) e − E T K D E − eT K ee
2 

 1 
= − E T K D E − eT  K e − αT  H ( x ) − C ( x , x )  e (3.23)
 2 

 1 
≤ − E T K D E − eT  λ min {K e} − αT  H ( x ) − C ( x , x )  e
  2 

Note that in the derivation of (3.23), because ( 12 H ( x ) − C ( x , x )) is skew-symmetric,

it follows that eT ( 12 H ( x ) − C ( x , x ))e = 0 .
Because both H ( x ) and C ( x , x ) satisfy the Lipschitz condition (Filippov 1988),
one has

1
αT  H ( x ) − C ( x,x ) ≤ c1  x + c2  x 
2 

where c1 and c2 are Lipschitz constants. Then (3.23) becomes

V ≤ − E T K D E − eT (λ min {K e} − (c1  x  + c2  x ))e (3.24)

If the control gain matrix K e is chosen to satisfy λ min {K e} = C1  x  +C2  x  ≥

c1  x  + c2  x  , V ≤ 0 . Therefore, E and e are bounded in terms of L2 norm
because they appear in V . From (3.18), it is known that x is bounded, and then
36 Synchronization and Control of Multiagent Systems

from (3.15), e is bounded. From (3.14), E is further bounded. As a result, E
and e are uniformly continuous; hence, based on Barbalat’s lemma, E → 0 and
e → 0 as time t → ∞. This implies that x → 0 and x → 0 as time t → ∞. From
the error dynamics (3.18), it is further concluded that there exists an invariant set
Ψ = {( x,x ) : x = 0, E = 0, e = 0, ε = 0} . Therefore, LaSalle’s theorem directly implies
asymptotic stability of the system (i.e., E → 0 , e → 0, and ε → 0 as t → ∞).
Theorem 3.1 holds provided that the following condition is satisfied:

1
λ min {K e} ≥ αT  H ( x ) − C ( x,x )
2 

or

λ min {K e} = C1  x  +C2  x  ≥ c1  x  + c2  x 

To determine the control gain K e in practical implementation, it is not necessary

to know H ( x ) and C ( x,x ) exactly in selection of C1 and C2. Instead, the control
gain K e (or C1 and C2 ) can be chosen large enough to meet this condition. In this
way, the controller will not be model dependent.
The following strategy can be used to guide the control gain tuning:

1. First, select α = 0 and K e = 0, and tune K P and K D using the trial-and-

error method. At this moment, the control law is equivalent to a standard
PD control without coupling.
2. Second, increase the value of α to add the synchronization control grad-
ually. As is seen from (3.13), the value of α determines the weight of
synchronization error ε in coupled position error E, and thus affects the
percentage of synchronization effort in the entire control action.
3. Increase K e from zero to ensure λ min {K e} ≥ αT ( 12 H ( x ) − C ( x , x ))  . The
value of K e depends on the value of α as well as the motion coordinates x
and x based on the Lipschitz condition.
4. After doing the above, K P and K D may need further adjustment using the
trial-and-error method.

Remark
Although the control law (3.16) can be expressed in a PD error feedback formulation
by submitting (3.14) into (3.16)—that is,

τ = K P ( I + αT )e + [ K D ( I + αT ) + ( I + αT )−1 K e ]e

= K Pe + K De (3.25)
Model-Free Synchronization Control of Multiple Motion Axes 37

it differs from the standard PD control in that the controller (3.16) explicitly drives
both the position error e and the coupled position error E to zero simultaneously,
and the standard PD control considers only convergence of the position error e. In
the synchronous controller (3.16), the synchronization constraint among multiple
axes, expressed in matrix T, is added to the control loop, and the control gains K P
and K D corresponding to e and e are chosen by following the regulation as shown
in (3.25). As a result, convergence of the synchronization error ε to zero is also
guaranteed during the motion, which leads to satisfactory transient performance of
synchronization.

3.4 Multiaxis Synchronization in Tracking Control

In the last section, a PD-based model-free synchronization control algorithm is
introduced to synchronize multiaxis motions with setpoint position control scheme.
In most practical applications, however, a tracking control between the initial and
final positions instead of setpoint position control is more often utilized. This is
because the moving velocity is also regulated to follow the desired value in the track-
ing control, which results in a better motion control effect. Tracking control strategy
has been widely utilized in many commercial motion control systems. In this sec-
tion, extending the setpoint position control as introduced in Section 3.3, a tracking
controller will be introduced to synchronize multiaxis motions.
Unlike setpoint position control, the actual velocity x is required to follow the
desired velocity x d during the motion with the tracking control. The desired veloc-
ity x d is assumed to be bounded up to its second time derivative. In many industrial
applications, the desired velocity at the final desired position is zero (i.e., x d = 0).
This case was called “setpoint tracking” by Sun et al. (2007).
The synchronous tracking controller is designed by adding feedforward controls
with x d and xd and a saturation function to the setpoint position control law (3.16),
expressed by

τ = K H xd + K C x d + K P E + K D E + ( I + αT )−1 K ee + sign( E ) K N (3.26)

where K H and KC are positive feedforward control gains, and K N is a parameter

that meets the following condition:
(3.27)
K N = ∆ H || xd || + ∆ C || x d ||

in which ∆ H and ∆ C are scalars. To make the controller independent in the model,
the control gains K H and KC are utilized instead of the modeling parameters H ( x )
and C ( x , x ) . The last term in (3.26) is a saturation function used to compensate for
the effect caused by the difference between the control gains K H and KC and the
real modeling parameters H ( x ) and C ( x , x ) .
Exploring the Variety of Random
Documents with Different Content
The Project Gutenberg eBook of Gangway for
Homer
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: Gangway for Homer

Author: George R. Hahn

Illustrator: John R. Forte

Release date: March 21, 2024 [eBook #73220]

Language: English

Original publication: New York, NY: Columbia Publications Inc, 1942

Credits: Greg Weeks, Mary Meehan and the Online Distributed

Proofreading Team at http://www.pgdp.net

*** START OF THE PROJECT GUTENBERG EBOOK GANGWAY

FOR HOMER ***
 GANGWAY FOR HOMER

By GEORGE R. HAHN

Illustrated by John Forte

C'mon out of the shadows, Homer. Here's one who claims you
as his patron. Unstring your lyre, mighty bard and sing
the epic of Achilles Maravain, who can't be hurt by bullets,
bombs, or blasters, and whose touch brings instant death!

[Transcriber's Note: This etext was produced from

Science Fiction Quarterly Spring 1942.
Extensive research did not uncover any evidence that
the U.S. copyright on this publication was renewed.]
His name was really John Smith. Incredibly enough, it had always
been John Smith. As far back as people in his circle and
neighborhood could remember, it had been John Smith—and they
could remember back all the way to when he had been a mere
tottering tot—to the swaddling clothes days. He was what might be
called a medium man. His height was medium. His middle-age was
medium. His hair, eyes, and nose were medium. Unpretentious he
looked and adequate. He fulfilled his name, which, as we mentioned
above, was John Smith—not Achilles Maravain.
Yet she persisted in calling him Achilles Maravain. She declaimed;
she cried out; she excited herself and all present—all to the effect
that John Smith was Achilles Maravain. Everybody paid her the best
of attention, although they couldn't believe her. Everybody regarded
her with interest. She had a wild, pale, exotic-looking face, a figure it
would be indelicate to remark upon, and legs that were crystallized
ecstasy. They listened to her words; they gazed upon her. The Los
Angeles Forum of Camera Arts gazed to satiety on face and figure
and legs and sighed en masse. Insanely gay sighs, sighed they. She
was desirable and moreover she had interrupted President Soupy's
discussion of "Repentance," a camera study in monotone by Pierre
de la Bardier. Had you ever listened to President Soupy remarking
that such and such was "taken on a Zeiss Super Iconta B with
Metchnormatic Ultra-Lite film at an exposure of f5.6 with diaphragm
—, etc.," you too would have been gay—aye, insanely gay—to have
had him interrupted by a luscious looking pair of legs like that. Thus,
everybody was happy. The LAFCA was happy; the lost-looking
scientist with the galvanometer and other trivia was happy, and the
three dapper young men were happy. Even John Smith was happy.
His happiness was obvious in the reluctance with which he took his
departure, in his formality. He rose from his seat and said to the
three dapper young men, of the hard, virtuous faces, "The Federal
Bureau of Investigation, I presume?"
"Right," crisply, youngishly. "John Smith, alias Achilles Maravain, you
are under arrest on charges of murder, seditious conduct, and high
treason against the government of the United States. Will you come
along quietly?"
Achilles—for it is as such he is to be referred to forever hence—did
not come along quietly. He did not come along. In point of fact, he
went his own way. The three grim young men of the FBI bitterly
contested his going, and, since, as everyone knows nowadays, to
touch Achilles Maravain was to undergo collapse, disintegration, and
death, the results were unfortunate.

This was the first and perhaps most important incident in his history.
It was the acorn from which sprouted that large and aberrant oak-
tree that was Achilles Maravain.
The next important incident—a scene perhaps even more diverting
than the last—was the Lincoln Heights scene. As the odds are
against it that the reader of this is either an archeologist or some
pervertedly informed devotee of ancient Los Angeles topography, it
is excusable to mention that Lincoln Heights was the jail of the city,
an institution comparable in purpose to our modern concentration
camps, but differing in that it was merely a squat, few-story cement
structure abundantly furnished with steel bars, locks, chains, gyves,
paraphernalia, and policemen. Its architecture was thus ideal for
Achilles' purposes. His purposes being to imprison the prison,
purposes in which he succeeded.
His remarkable feat first manifested itself when Sergeant Leery
crashed titanically into nothingness. Not actual nothingness—as was
evidenced by its palpability—but a substance that, for all practical
purposes, was nonexistent; all practical purposes that is except that
of preventing exit or entry in regards to the Lincoln Heights jail.
Sergeant Leery withdrew his nose a few paces, vigorously rubbing
that injured member, and stared quizzically at this absurd tangibility.
He stared for a long and ponderous time and then began shouting.
Minions of the law popped miraculously into view at this point, as if
conjured there by the magic of Leery's stentorian voice. Miraculously
they popped and popping, equally miraculously popped no more.
The invisible barrier restrained them; it framed their popping faces,
their popping eyes. It kept them within the building, sealing the
doors, the windows, the walls. It was, in fact a prison; Achilles
Maravain had imprisoned the prison.
Had he stopped there, there's a shade of a ghost's super-attenuated
chance that all might have been forgotten, except perhaps by the
infuriated gendarmerie and prisoners who were left permanently to
their own devices within the Lincoln Heights jail. But Achilles didn't
stop. He visited the First Street Headquarters Jail and imprisoned it.
He visited all the jails. Likewise the insane asylums and the
hospitals. Personal appearance tour, it was; an interstate tour. He
went to Salt Lake City and there gave a repeat performance. Jails,
hospitals, etc. Thence to Denver; thence to Topeka; thence to
Kansas City. Followed by St. Louis. Followed by Indianapolis. And
on all the way to the East Coast.
It is not to be supposed that he was uncontested in this progression.
Very much to the contrary. He was shot at. Often and with the utmost
accuracy was he fired upon. Apparently, however, with no effect for
he seemed invulnerable.
Not elusive was our Achilles, not wily, not adroit. Not even clever. He
was merely invulnerable and clumsily so to boot. He would wade into
a mass of stalwart police or soldiers—the militia tried cannons on
him—and projectiles would simply bounce away from him. They
would explode in the conventional manner. Only no fragments or
concussion waves apparently could reach him. After this, the
opposition would be scattered like the proverbial chaff.
It was a melee, a very horrifying and immensely entertaining mess.
Chaos there was and wildness and fantasy and even fanaticism.
Yes, even the latter. This last was instanced the time a group of
misdirected fans of Achilles misconstrued him, and, in the belief that
he was the Almighty, surrounded him in the midst of the pursuit of
one of his more stirring enterprises. He misconstrued them, too.
They still remain, so far as is known, in the housing he provided their
zeal.

Now we return to the beginning of the story. Not for the sake of
confusion, but merely to pick up a most important thread. Remember
the Camera Forum scene? And Los Angeles? Los Angeles, if you
are following the mood of this story, is mere dust and collapse by
now. Nevertheless, we return to the vanished metropolis and to the
Forum, the three young men of the FBI, and to the lost-looking
scientist with the galvanometer and other trivia—the scientist whom
you probably never noticed, having been lost in the spell of her. The
lost-looking scientist was happy, too.
His happiness lay in that he had come to a conclusion, one affecting
Achilles Maravain. His conclusion was that Maravain was
scientifically explainable. Not just his feats; not just the decimation
he wrought upon police; not just the prisons in which he enveloped
prisons. No—more than that—the works. Everything about Achilles
Maravain—his personality; his attitude toward life, love and literature
—all down to his very kneejerks.
First and most important of all, our Achilles had an inferiority
complex. Definitely. The proposition that anyone who had actually,
with reason, been called John Smith all his life did not have an
inferiority complex was fantastic. But the man's actions proved it
beyond doubt: he picked on criminals, insane, and the sick because
he felt inferior to them, and compensated thus. Amazing logic? Well,
everyone thought so at the time, although as you can see, it was
really extraordinarily simple.
But, at the time, everyone was amazed, even the scientist himself.
He gloried in it, glowed and, entering further into the spirit and tempo
of his theories, babbled out point after telling point. Argued.
Philosophized. He quoted statistics about the ratio of invention to the
inferiority-complex and compared it with the results Achilles had
obtained. He proved that ultra-vibrational force-walls—this being
essentially what Achilles had developed for the demolition of law and
order and for the production of honestagawd, fool-proof, tamper-
proof prisons—were Machiavellian, Mephistophelian, and just plain
hellish. Why invent them, then, except to demonstrate a superiority
the inventor really didn't feel?
The scientist meditated further, brooded, calculated, grunted awhile
and then predicted—or, as he put it: prognosticated—that Achilles
would declare himself a dictator.
Which Achilles did.
In this, however, there was a flaw; here lay his weakness. Not in the
actual fact that he protested himself the greatest and wisest of men,
but that he attempted wiles. He didn't come out with it forthrightly; he
wasn't blunt as he had been with his interesting massacres. He
proved himself cagey, contemptible, striking the Humanitarian pose.
He was, he stated, producing all these absorbing newspaper stories
for man's own good. Or, rather, Man's. Man with a capital M. A
document he issued, long and scholarly. It reeked; it stank; it was
crawling with hypocrisy and shoddy diplomacy. He took some thirty
thousand words to indicate that pestilence, famine, and war was in
existence. That thieving, murder, and kindred rot was also in
evidence. He dithered about the general theme that this was
horrible. Tediously he pedanted, hedging around concerning the
Perfect State, eventually coming out into the open with his own
private Perfect State plan. Revised and condensed it still reeked. Get
rid of all the misfits and criminals and the insane. Prison up the
squarepegs and breed them out. And then direct democracy just as
the Greeks had.
Apparently he had never heard of economics. No one had told him
that Greek democracy existed on the basis of a slave system. No
one had told him of other things that had either been thought of,
worked out, or had evolved according to the scientific laws
concerning economics and society since the time of the Greeks.
Achilles Maravain was stuck on Homeric Greek democracy—only he
indicated that he, personally, would be Democrat Number One.
Again we bewilder the reader with a thread from the beginning.
Again we return to the Los Angeles Camera Forum Scene. This time
to call to the mind of the elated reader that succulent item of
femininity that first claimed our attention with her sprightly
uncovering of Achilles Maravain as the seemingly innocuous John
Smith.
We find Cecile Douve, as she is known to the intelligence services of
this and perhaps a few other countries, in a stinky little bedroom.
Again don't get ahead of the story; she is merely investigating. Not
engaged in active inquiry, if you follow me.
This stinky little bedroom, with massive volumes of a technical
nature, broken test tubes, and other rot and junk of a like nature
littering it, is the erstwhile bedroom of Achilles Maravain. He no
longer inhabits it, although we can linger nostalgically for a moment,
although we can sniff mystifiedly at the—peculiar—odors emanating
from the broken test tubes, although we can tinker with the
gimcracks and thingumbobs and machinery and no doubt shock
ourselves into a reckoning with Old Scratch.
In any case, Cecile Douve is here searching for a clue to the
whereabouts of Achilles. The scientist of the galvanometers is also
here. His name is Harold Boscoe, and he is a Ph. D. Together,
Cecile and the Ph. D. search and also engage in polite converse.
They sniff not, mystifiedly or otherwise; they linger not on anything
nor brood about the fact that perhaps here, in this very, very room
was conceived the diabolicism of the force-wall. No, they search and
converse.
"It must not happen. The man is a maniac," postulates Cecile prettily,
then continues the efficient search.
"Honeybunch,"—evidently the poor egg has joined the clan of the
lovelorn—"it shall not. I shall find something to combat him and his
evil."
"Do you think you can do it, my pet?"
"Certainly. I'm a scientist, am I not? Just between you and me (and a
few governments: Auth. Note) I'm working on something already. I
have a magnificent conception that may well prove his downfall."
"Do you really think so? You're so wise—so—so marvelous."
"Do you really think so?"
"Yes," a pause, then in husky tones, "really."
Embraces, osculations, and speeches. At precisely the right
moment, when his devotion is white hot, she molds him and sends
him back to work.

At approximately this same time, there is going on a very important

meeting of various high and significant officials of the government:
the President, Vice President, Cabinet Members, House Committee
on Achilles Maravain, Senatorial Investigation Committee on Achilles
Maravain, the current successor to the Dies Committee, and the First
Lady. Hubbub, clamor, chaos. The authoritative voice of the
President lifts.
"Silence, please."
The Chairman of the House Committee on Achilles Maravain rises
and addresses the President. "Mr. President," (cough), "we are led to
believe that these are the facts of the case. There is (or are) a
person (or persons) calling himself (or themselves) Achilles
Maravain, who has been (or have been) imprisoning prisons,
causing no end of annoyance and embarrassment, and who has (or
have) proclaimed what purports to be the constitution of a new
American state, founded after the manner of our Grecian
predecessors in the experiment of democracy. The experiment of
democracy, which, may I say, gentlemen" (voice takes oratorical
tones) "has fulfilled all the most rosy hopes and expectations, which
has turned a barren wilderness of thirteen original states into the
magnificence and resplendency of—"
"We are all aware of that, Mr. Ainsworth." The President's grimace
could rightly be termed sinister.
The Chairman sits down abruptly and the President continues. "In
any event, gentlemen, we are confronted by a profoundly serious
situation, coming as it is, at a time when we should bend every effort
toward preparing for a war against the Old World. This person (or
persons) known as Achilles Maravain is having a distressingly
diverting effect upon us when all energies from that—er, madman—
in Europe. Beyond doubt he is some pseudo-idealistic radical—
perhaps an emissary of the Federation. Whatever he is, he has no
full cognizance of the extreme gravity of our situation, and, as such
will not assist us. Thus, he (or they) must be destroyed. What might
any of you gentlemen suggest towards the speedy expedition of this
destruction?"
The First Lady arises. "I'm not a gentleman," she simpers, "but—"
"Quite true. Sit down, please."
"Wait a moment. I have something to suggest. Perhaps you do not
realize it, but I am indirectly responsible for the uncovering of
Achilles Maravain, in Smith's clothing. It was one of my girls, Cecile
Douve, who did the job. And we—she in active duty, and I as her
patron—are continuing our efforts. As once she wormed her way into
his affections, so shall she do it again. As once she effectively
uncovered him, so shall she do a second time. I really promise you
results, gentlemen. Results."

And results she got. Results they were. The web, the power
expended, the intricacies of thought, the drive of five hundred
individuals were her results. The huge rolling mass of energy that
was exhausted by five hundred highly specialized and superbly
trained and educated beings was the result she directed against the
insidious Mr. Maravain. And, most important of all, one person
named Cecile Douve. Four hundred and ninety-nine engineers,
scientists, technicians, and one little lump of hotcha generally known
as Cecile Douve.
"I love you," she said.
Achilles replied. "The last time I believed that, you called in the FBI
as witnesses to our mutual affection."
"I was mad, my darling. I didn't understand you." (Hushed, reverent
tones.) "Even then I felt violently attracted to you, to you as a man,
but your purposes and powers seemed so fearful ... I thought you
were a madman and myself a monster to love you. But now I know
... when I read your wonderful proclamation, I realized how wrong I
had been—how gentle and idealistic you are. I understood then your
purity and realized the nobility of your aspirations.
"I love you."
She moved in for the clinch.
"It would only be fair to warn you," he replied, "that I still have the
force screen armoring me. Cuddling under these conditions would be
quite inadvisable."
She recoiled in a somewhat unamorous fashion.
"Still," he continued, "I love you, too. I don't want to trust you—but I
do. Don't look hopeful my dear—I don't completely. Just to a certain,
reasonable degree. So, here's what: if my noble aspirations pan out,
as I can't help but expect they will, I'll marry you. In the meantime,
we can be friends. We can conduct a pleasant, frolicky little
association, however—an entirely platonic one."
He sighed. "Would that Homer were alive today to write the story of
Achilles Maravain as it should be written. Will I have poets worthy of
me?"
This, she thought, could go on indefinitely. "How soon will it be?" she
broke in. "When will you succeed? Can you make it very soon, my
dearest?"
"It can't be any too soon for me, either, dewdrop—but restrain
yourself."
"I can't—oh, I can't!" she cried. Heavy breathing, then, in more
serious tones. "I know what to do. I have influence in Washington. I'll
arrange an audience for you with the President. With the President
and all of Congress. They'll see you."
"Nice of them, but I don't see the use of it."
"It might be of inestimable use, my darling. A direct impact of your
personality and honesty and drive should convince them. It would be
almost certain to convince them; they're only human, my dear. And
think of the time and trouble we can save if they are ready to give in
gracefully. Please!"
"Very well," he sighed. "I'll do it. Don't think for a moment I don't
suspect treachery, my pet, but after all, I am invincible. You know
that, I hope."

And on this note ends the reconciliation. Immediately followed by

much ado. Preparations while four hundred and ninety-nine
engineers, etc., work in a veritable frenzy. And, out of their efforts
and energy, there grows an amphitheatre, large and capacious.
Pretty and modern. Beautiful.
This was to be the scene of the meeting. Here is to be decided the
fate of more than a hundred and thirty million people. Here is to be
expounded the rules and laws of a state founded on Grecian lines,
on the classic examples.
Here, on March 15.
Beware of the Ides of March, O Caesar.
Glorious, powerful, invincible Achilles Maravain comes to the
amphitheatre. Nowadays with the details and obscurities of the
episode in history shuffled into relative inconspicuity, one doesn't
know precisely how the cards fell or get the subtleties of the deal.
Did any soothsayers annoy our equivalent of Caesar on his route;
did his nonexistent Calpurnia dream gorily the night before; did
lionesses whelp in the streets, or did fierce, fiery warriors fight upon
the clouds "which drizzled blood upon the capital"? Gangway for
Homer, or even Shakespeare. Either of these two could have done
justice to our play.
In any case, Achilles ignored whatever omens there might have
been and came to the amphitheatre on March 15.
Cecile met him at the great, modern-looking portal and led him in,
introducing him to her benefactress, the First Lady, who, in turn
consummated the formalities with the President himself. Achilles was
very well-behaved throughout these presentations, conducting
himself with decorum and consideration for all the people who eyed
his much-publicized armor with especial dubiety. He was very
pleased with himself about the whole thing. All these key figures,
these obstructions to his philosophy, this destructible humanity,
ponderous, ripe so to speak for explosions and force walls—and he
showing such admirable restraint about it all. Indeed, he felt content.
Restraint, control, self-discipline—these his watch-words.
 She introduced him to the First Lady.

The President didn't take him by the hand, the force wall preventing
—but he did the next best thing. He preceded him to the raised dais
in the centre of the amphitheatre and, from the spot, delivered a
fetching little introduction about which no more severe a criticism can
be applied than "superfluous." After this, Achilles began his talk.
Here also is the ubiquitous scientist of the lost-looking face.
Apparently a member, if not a chieftain, of the clan of the four
hundred and ninety-nine technicians. He is looking remarkably
heroic at the moment. Almost gigantic—in a spiritual sort of way.
He turns and throws a switch.
And, in the amphitheatre, a globular hemisphere descends upon the
dais supporting Achilles Maravain, immediately transforming him into
a raging Achilles. A half-spheroid, transparent, glassy, but
immensely malleable and tensile and strong.

Upstairs in the little room in which stands the heroic and lost-looking
scientist, the door flies open. Cecile Douve, betrayer extraordinary,
hotcha extraordinary, flies into his arms.
"Darling, the hemisphere is cracking—he's winning out. What'll he do
to me?" All this excitedly. Then, ruminatively, almost sadly. "He won't
want to marry me now."
"Never fear, my sweetness," replies the chieftain of the four hundred
and ninety-nine. "We will win out. Earth science shall triumph. The
hemisphere is just makeshift, to hold him in one spot for a minute or
two. Earth'll really get going in a sec. Earth is insuperable.
Classicism he wanted and classicism he'll get. Remember the first
Achilles? He had a vulnerable spot. His heel!" The lost look was
replaced by a malevolent grin, sage and content. "Achilles Maravain
has a heel, too. It couldn't be protected by the force-wall, could it?
He doesn't walk on an inch of apparent nothingness does he? No.
He's vulnerable, just as his Homeric predecessor. And we don't have
to use clumsy poisoned arrows on this"—sneering emphasis
—"heel." A wild laugh. "We just throw a shot of good old electricity
into him."
On the dais, the violent, raging figure of John Smith, alias Achilles
Maravain, colossus of the classics, exponent of the ages, Caesar
omnipotent, stiffens convulsively as a couple of hundred thousand
volts of electricity crisps his flesh. For a long moment, what is left of
him remains upright. Then, quietly it falls.
Achilles number two seeks out his illustrious predecessor in Elysia.
 *** END OF THE PROJECT GUTENBERG EBOOK GANGWAY
FOR HOMER ***

Updated editions will replace the previous one—the old editions will
be renamed.

Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States copyright in
these works, so the Foundation (and you!) can copy and distribute it
in the United States without permission and without paying copyright
royalties. Special rules, set forth in the General Terms of Use part of
this license, apply to copying and distributing Project Gutenberg™
electronic works to protect the PROJECT GUTENBERG™ concept
and trademark. Project Gutenberg is a registered trademark, and
may not be used if you charge for an eBook, except by following the
terms of the trademark license, including paying royalties for use of
the Project Gutenberg trademark. If you do not charge anything for
copies of this eBook, complying with the trademark license is very
easy. You may use this eBook for nearly any purpose such as
creation of derivative works, reports, performances and research.
Project Gutenberg eBooks may be modified and printed and given
away—you may do practically ANYTHING in the United States with
eBooks not protected by U.S. copyright law. Redistribution is subject
to the trademark license, especially commercial redistribution.

START: FULL LICENSE

THE FULL PROJECT GUTENBERG LICENSE
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebooknice.com