Description
Book SynopsisCyberPhysicalHuman Systems A comprehensive edited volume exploring the latest in the interactions between cyberphysical systems and humans
In CyberPhysicalHuman Systems: Fundamentals and Applications, a team of distinguished researchers delivers a robust and up-to-date volume of contributions from leading researchers on CyberPhysicalHuman Systems, an emerging class of systems with increased interactions between cyberphysical, and human systems communicating with each other at various levels across space and time, so as to achieve desired performance related to human welfare, efficiency, and sustainability.
The editors have focused on papers that address the power of emerging CPHS disciplines, all of which feature humans as an active component during cyber and physical interactions. Articles that span fundamental concepts and methods to various applications in engineering sectors of transportation, robotics, and healthcare and general socio-technical system
Table of Contents
A Note from the Series Editor xvii
About the Editors xviii
List of Contributors xix
Introduction xxvii
Part I Fundamental Concepts and Methods 1
1 Human-in-the-Loop Control and Cyber–Physical–Human Systems: Applications and Categorization 3
Tariq Samad
1.1 Introduction 3
1.2 Cyber + Physical + Human 4
1.2.1 Cyberphysical Systems 5
1.2.2 Physical–Human Systems 6
1.2.3 Cyber–Human Systems 6
1.3 Categorizing Human-in-the-Loop Control Systems 6
1.3.1 Human-in-the-Plant 8
1.3.2 Human-in-the-Controller 8
1.3.3 Human–Machine Control Symbiosis 10
1.3.4 Humans-in-Multiagent-Loops 11
1.4 A Roadmap for Human-in-the-Loop Control 13
1.4.1 Self- and Human-Driven Cars on Urban Roads 13
1.4.2 Climate Change Mitigation and Smart Grids 14
1.5 Discussion 15
1.5.1 Other Ways of Classifying Human-in-the-Loop Control 15
1.5.2 Modeling Human Understanding and Decision-Making 16
1.5.3 Ethics and CPHS 18
1.6 Conclusions 19
Acknowledgments 19
References 20
2 Human Behavioral Models Using Utility Theory and Prospect Theory 25
Anuradha M. Annaswamy and Vineet Jagadeesan Nair
2.1 Introduction 25
2.2 Utility Theory 26
2.2.1 An Example 27
2.3 Prospect Theory 27
2.3.1 An Example: CPT Modeling for SRS 30
2.3.1.1 Detection of CPT Effects via Lotteries 32
2.3.2 Theoretical Implications of CPT 33
2.3.2.1 Implication I: Fourfold Pattern of Risk Attitudes 34
2.3.2.2 Implication II: Strong Risk Aversion Over Mixed Prospects 36
2.3.2.3 Implication III: Effects of Self-Reference 37
2.4 Summary and Conclusions 38
Acknowledgments 39
References 39
3 Social Diffusion Dynamics in Cyber–Physical–Human Systems 43
Lorenzo Zino and Ming Cao
3.1 Introduction 43
3.2 General Formalism for Social Diffusion in CPHS 45
3.2.1 Complex and Multiplex Networks 45
3.2.2 General Framework for Social Diffusion 46
3.2.3 Main Theoretical Approaches 48
3.3 Modeling Decision-Making 49
3.3.1 Pairwise Interaction Models 49
3.3.2 Linear Threshold Models 52
3.3.3 Game-Theoretic Models 53
3.4 Dynamics in CPHS 55
3.4.1 Social Diffusion in Multiplex Networks 56
3.4.2 Co-Evolutionary Social Dynamics 58
3.5 Ongoing Efforts Toward Controlling Social Diffusion and Future Challenges 62
Acknowledgments 63
References 63
4 Opportunities and Threats of Interactions Between Humans and Cyber–Physical Systems – Integration and Inclusion Approaches for Cphs 71
Frédéric Vanderhaegen and Victor Díaz Benito Jiménez
4.1 CPHS and Shared Control 72
4.2 “Tailor-made” Principles for Human–CPS Integration 73
4.3 “All-in-one” based Principles for Human–CPS Inclusion 74
4.4 Dissonances, Opportunities, and Threats in a CPHS 76
4.5 Examples of Opportunities and Threats 79
4.6 Conclusions 85
References 86
5 Enabling Human-Aware Autonomy Through Cognitive Modeling and Feedback Control 91
Neera Jain, Tahira Reid, Kumar Akash, Madeleine Yuh, and Jacob Hunter
5.1 Introduction 91
5.1.1 Important Cognitive Factors in HAI 92
5.1.2 Challenges with Existing CPHS Methods 93
5.1.3 How to Read This Chapter 95
5.2 Cognitive Modeling 95
5.2.1 Modeling Considerations 95
5.2.2 Cognitive Architectures 97
5.2.3 Computational Cognitive Models 98
5.2.3.1 ARMAV and Deterministic Linear Models 99
5.2.3.2 Dynamic Bayesian Models 99
5.2.3.3 Decision Analytical Models 100
5.2.3.4 POMDP Models 102
5.3 Study Design and Data Collection 103
5.3.1 Frame Research Questions and Identify Variables 104
5.3.2 Formulate Hypotheses or Determine the Data Needed 105
5.3.2.1 Hypothesis Testing Approach 105
5.3.2.2 Model Training Approach 105
5.3.3 Design Experiment and/or Study Scenario 107
5.3.3.1 Hypothesis Testing Approach 107
5.3.3.2 Model Training Approach 107
5.3.4 Conduct Pilot Studies and Get Initial Feedback; Do Preliminary Analysis 108
5.3.5 A Note about Institutional Review Boards and Recruiting Participants 109
5.4 Cognitive Feedback Control 109
5.4.1 Considerations for Feedback Control 110
5.4.2 Approaches 111
5.4.2.1 Heuristics-Based Planning 111
5.4.2.2 Measurement-Based Feedback 112
5.4.2.3 Goal-Oriented Feedback 112
5.4.2.4 Case Study 112
5.4.3 Evaluation Methods 113
5.5 Summary and Opportunities for Further Investigation 113
5.5.1 Model Generalizability and Adaptability 114
5.5.2 Measurement of Cognitive States 114
5.5.3 Human Subject Study Design 114
References 115
6 Shared Control with Human Trust and Workload Models 125
Murat Cubuktepe, Nils Jansen, and Ufuk Topcu
6.1 Introduction 125
6.1.1 Review of Shared Control Methods 126
6.1.2 Contribution and Approach 127
6.1.3 Review of IRL Methods Under Partial Information 128
6.1.3.1 Organization 129
6.2 Preliminaries 129
6.2.1 Markov Decision Processes 129
6.2.2 Partially Observable Markov Decision Processes 130
6.2.3 Specifications 130
6.3 Conceptual Description of Shared Control 131
6.4 Synthesis of the Autonomy Protocol 132
6.4.1 Strategy Blending 132
6.4.2 Solution to the Shared Control Synthesis Problem 133
6.4.2.1 Nonlinear Programming Formulation for POMDPs 133
6.4.2.2 Strategy Repair Using Sequential Convex Programming 134
6.4.3 Sequential Convex Programming Formulation 135
6.4.4 Linearizing Nonconvex Problem 135
6.4.4.1 Linearizing Nonconvex Constraints and Adding Slack Variables 135
6.4.4.2 Trust Region Constraints 136
6.4.4.3 Complete Algorithm 136
6.4.4.4 Additional Specifications 136
6.4.4.5 Additional Measures 137
6.5 Numerical Examples 137
6.5.1 Modeling Robot Dynamics as POMDPs 138
6.5.2 Generating Human Demonstrations 138
6.5.3 Learning a Human Strategy 139
6.5.4 Task Specification 139
6.5.5 Results 140
6.6 Conclusion 140
Acknowledgments 140
References 140
7 Parallel Intelligence for CPHS: An ACP Approach 145
Xiao Wang, Jing Yang, Xiaoshuang Li, and Fei-Yue Wang
7.1 Background and Motivation 145
7.2 Early Development in China 147
7.3 Key Elements and Framework 149
7.4 Operation and Process 151
7.4.1 Construction of Artificial Systems 152
7.4.2 Computational Experiments in Parallel Intelligent Systems 152
7.4.3 Closed-Loop Optimization Based on Parallel Execution 153
7.5 Applications 153
7.5.1 Parallel Control and Intelligent Control 154
7.5.2 Parallel Robotics and Parallel Manufacturing 156
7.5.3 Parallel Management and Intelligent Organizations 157
7.5.4 Parallel Medicine and Smart Healthcare 158
7.5.5 Parallel Ecology and Parallel Societies 160
7.5.6 Parallel Economic Systems and Social Computing 161
7.5.7 Parallel Military Systems 163
7.5.8 Parallel Cognition and Parallel Philosophy 164
7.6 Conclusion and Prospect 165
References 165
Part II Transportation 171
8 Regularities of Human Operator Behavior and Its Modeling 173
Aleksandr V. Efremov
8.1 Introduction 173
8.2 The Key Variables in Man–Machine Systems 174
8.3 Human Responses 177
8.4 Regularities of Man–Machine System in Manual Control 180
8.4.1 Man–Machine System in Single-loop Compensatory System 180
8.4.2 Man–Machine System in Multiloop, Multichannel, and Multimodal Tasks 185
8.4.2.1 Man–Machine System in the Multiloop Tracking Task 185
8.4.2.2 Man–Machine System in the Multichannel Tracking Task 187
8.4.2.3 Man–Machine System in Multimodal Tracking Tasks 188
8.4.2.4 Human Operator Behavior in Pursuit and Preview Tracking Tasks 191
8.5 Mathematical Modeling of Human Operator Behavior in Manual Control Task 194
8.5.1 McRuer’s Model for the Pilot Describing Function 194
8.5.1.1 Single-Loop Compensatory Model 194
8.5.1.2 Multiloop and Multimodal Compensatory Model 197
8.5.2 Structural Human Operator Model 197
8.5.3 Pilot Optimal Control Model 199
8.5.4 Pilot Models in Preview and Pursuit Tracking Tasks 201
8.6 Applications of the Man–Machine System Approach 202
8.6.1 Development of Criteria for Flying Qualities and PIO Prediction 203
8.6.1.1 Criteria of FQ and PIO Prediction as a Requirement for the Parameters of the Pilot-Aircraft System 203
8.6.1.2 Calculated Piloting Rating of FQ as the Criteria 205
8.6.2 Interfaces Design 206
8.6.3 Optimization of Control System and Vehicle Dynamics Parameters 210
8.7 Future Research Challenges and Visions 213
8.8 Conclusion 214
References 215
9 Safe Shared Control Between Pilots and Autopilots in the Face of Anomalies 219
Emre Eraslan, Yildiray Yildiz, and Anuradha M. Annaswamy
9.1 Introduction 219
9.2 Shared Control Architectures: A Taxonomy 221
9.3 Recent Research Results 222
9.3.1 Autopilot 224
9.3.1.1 Dynamic Model of the Aircraft 224
9.3.1.2 Advanced Autopilot Based on Adaptive Control 225
9.3.1.3 Autopilot Based on Proportional Derivative Control 228
9.3.2 Human Pilot 228
9.3.2.1 Pilot Models in the Absence of Anomaly 228
9.3.2.2 Pilot Models in the Presence of Anomaly 229
9.3.3 Shared Control 230
9.3.3.1 SCA1: A Pilot with a CfM-Based Perception and a Fixed-Gain Autopilot 231
9.3.3.2 SCA2: A Pilot with a CfM-Based Decision-Making and an Advanced Adaptive Autopilot 232
9.3.4 Validation with Human-in-the-Loop Simulations 232
9.3.5 Validation of Shared Control Architecture 1 234
9.3.5.1 Experimental Setup 234
9.3.5.2 Anomaly 235
9.3.5.3 Experimental Procedure 235
9.3.5.4 Details of the Human Subjects 236
9.3.5.5 Pilot-Model Parameters 237
9.3.5.6 Results and Observations 237
9.3.6 Validation of Shared Control Architecture 2 240
9.3.6.1 Experimental Setup 241
9.3.6.2 Anomaly 241
9.3.6.3 Experimental Procedure 242
9.3.6.4 Details of the Human Subjects 243
9.3.6.5 Results and Observations 244
9.4 Summary and Future Work 246
References 247
10 Safe Teleoperation of Connected and Automated Vehicles 251
Frank J. Jiang, Jonas Mårtensson, and Karl H. Johansson
10.1 Introduction 251
10.2 Safe Teleoperation 254
10.2.1 The Advent of 5G 258
10.3 CPHS Design Challenges in Safe Teleoperation 259
10.4 Recent Research Advances 261
10.4.1 Enhancing Operator Perception 261
10.4.2 Safe Shared Autonomy 264
10.5 Future Research Challenges 267
10.5.1 Full Utilization of V2X Networks 267
10.5.2 Mixed Autonomy Traffic Modeling 268
10.5.3 5G Experimentation 268
10.6 Conclusions 269
References 270
11 Charging Behavior of Electric Vehicles 273
Qing-Shan Jia and Teng Long
11.1 History, Challenges, and Opportunities 274
11.1.1 The History and Status Quo of EVs 274
11.1.2 The Current Challenge 276
11.1.3 The Opportunities 277
11.2 Data Sets and Problem Modeling 278
11.2.1 Data Sets of EV Charging Behavior 278
11.2.1.1 Trend Data Sets 279
11.2.1.2 Driving Data Sets 279
11.2.1.3 Battery Data Sets 279
11.2.1.4 Charging Data Sets 279
11.2.2 Problem Modeling 281
11.3 Control and Optimization Methods 284
11.3.1 The Difficulty of the Control and Optimization 284
11.3.2 Charging Location Selection and Routing Optimization 285
11.3.3 Charging Process Control 286
11.3.4 Control and Optimization Framework 287
11.3.4.1 Centralized Optimization 287
11.3.4.2 Decentralized Optimization 288
11.3.4.3 Hierarchical Optimization 288
11.3.5 The Impact of Human Behaviors 289
11.4 Conclusion and Discussion 289
References 290
Part III Robotics 299
12 Trust-Triggered Robot–Human Handovers Using Kinematic Redundancy for Collaborative Assembly in Flexible Manufacturing 301
S. M. Mizanoor Rahman, Behzad Sadrfaridpour, Ian D. Walker, and Yue Wang
12.1 Introduction 301
12.2 The Task Context and the Handover 303
12.3 The Underlying Trust Model 304
12.4 Trust-Based Handover Motion Planning Algorithm 305
12.4.1 The Overall Motion Planning Strategy 305
12.4.2 Manipulator Kinematics and Kinetics Models 305
12.4.3 Dynamic Impact Ellipsoid 306
12.4.4 The Novel Motion Control Approach 307
12.4.5 Illustration of the Novel Algorithm 308
12.5 Development of the Experimental Settings 310
12.5.1 Experimental Setup 310
12.5.1.1 Type I: Center Console Assembly 310
12.5.1.2 Type II: Hose Assembly 311
12.5.2 Real-Time Measurement and Display of Trust 311
12.5.2.1 Type I: Center Console Assembly 311
12.5.2.2 Type II: Hose Assembly 313
12.5.2.3 Trust Computation 313
12.5.3 Plans to Execute the Trust-Triggered Handover Strategy 314
12.5.3.1 Type I Assembly 314
12.5.3.2 Type II Assembly 314
12.6 Evaluation of the Motion Planning Algorithm 315
12.6.1 Objective 315
12.6.2 Experiment Design 315
12.6.3 Evaluation Scheme 315
12.6.4 Subjects 316
12.6.5 Experimental Procedures 316
12.6.5.1 Type I Assembly 317
12.6.5.2 Type II Assembly 317
12.7 Results and Analyses, Type I Assembly 318
12.8 Results and Analyses, Type II Assembly 322
12.9 Conclusions and Future Work 323
Acknowledgment 324
References 324
13 Fusing Electrical Stimulation and Wearable Robots with Humans to Restore and Enhance Mobility 329
Thomas Schauer, Eduard Fosch-Villaronga, and Juan C. Moreno
13.1 Introduction 329
13.1.1 Functional Electrical Stimulation 330
13.1.2 Spinal Cord Stimulation 331
13.1.3 Wearable Robotics (WR) 332
13.1.4 Fusing FES/SCS and Wearable Robotics 334
13.2 Control Challenges 335
13.2.1 Feedback Approaches to Promote Volition 336
13.2.2 Principles of Assist-as-Needed 336
13.2.3 Tracking Control Problem Formulation 336
13.2.4 Co-operative Control Strategies 337
13.2.5 EMG- and MMG-Based Assessment of Muscle Activation 344
13.3 Examples 345
13.3.1 A Hybrid Robotic System for Arm Training of Stroke Survivors 345
13.3.2 First Certified Hybrid Robotic Exoskeleton for Gait Rehabilitation Settings 347
13.3.3 Body Weight-Supported Robotic Gait Training with tSCS 348
13.3.4 Modular FES and Wearable Robots to Customize Hybrid Solutions 348
13.4 Transfer into Daily Practice: Integrating Ethical, Legal, and Societal Aspects into the Design 350
13.5 Summary and Outlook 352
Acknowledgments 353
Acronyms 353
References 354
14 Contemporary Issues and Advances in Human–Robot Collaborations 365
Takeshi Hatanaka, Junya Yamauchi, Masayuki Fujita, and Hiroyuki Handa
14.1 Overview of Human–Robot Collaborations 365
14.1.1 Task Architecture 366
14.1.2 Human–Robot Team Formation 368
14.1.3 Human Modeling: Control and Decision 369
14.1.4 Human Modeling: Other Human Factors 371
14.1.5 Industrial Perspective 372
14.1.6 What Is in This Chapter 375
14.2 Passivity-Based Human-Enabled Multirobot Navigation 376
14.2.1 Architecture Design 377
14.2.2 Human Passivity Analysis 379
14.2.3 Human Workload Analysis 381
14.3 Operation Support with Variable Autonomy via Gaussian Process 383
14.3.1 Design of the Operation Support System with Variable Autonomy 385
14.3.2 User Study 388
14.3.2.1 Operational Verification 388
14.3.2.2 Usability Test 390
14.4 Summary 391
Acknowledgments 393
References 393
Part IV Healthcare 401
15 Overview and Perspectives on the Assessment and Mitigation of Cognitive Fatigue in Operational Settings 403
Mike Salomone, Michel Audiffren, and Bruno Berberian
15.1 Introduction 403
15.2 Cognitive Fatigue 404
15.2.1 Definition 404
15.2.2 Origin of Cognitive Fatigue 404
15.2.3 Effects on Adaptive Capacities 406
15.3 Cyber–Physical System and Cognitive Fatigue: More Automation Does Not Imply Less Cognitive Fatigue 406
15.4 Assessing Cognitive Fatigue 409
15.4.1 Subjective Measures 409
15.4.2 Behavioral Measures 410
15.4.3 Physiological Measurements 410
15.5 Limitations and Benefits of These Measures 412
15.6 Current and Future Solutions and Countermeasures 412
15.6.1 Physiological Computing: Toward Real-Time Detection and Adaptation 412
15.7 System Design and Explainability 414
15.8 Future Challenges 415
15.8.1 Generalizing the Results Observed in the Laboratory to Ecological Situations 415
15.8.2 Determining the Specificity of Cognitive Fatigue 415
15.8.3 Recovering from Cognitive Fatigue 417
15.9 Conclusion 418
References 419
16 Epidemics Spread Over Networks: Influence of Infrastructure and Opinions 429
Baike She, Sebin Gracy, Shreyas Sundaram, Henrik Sandberg, Karl H. Johansson, andPhilipE.Paré
16.1 Introduction 429
16.1.1 Infectious Diseases 429
16.1.2 Modeling Epidemic Spreading Processes 430
16.1.3 Susceptible–Infected–Susceptible (SIS) Compartmental Models 431
16.2 Epidemics on Networks 432
16.2.1 Motivation 432
16.2.2 Modeling Epidemics over Networks 433
16.2.3 Networked Susceptible–Infected–Susceptible Epidemic Models 434
16.3 Epidemics and Cyber–Physical–Human Systems 436
16.3.1 Epidemic and Opinion Spreading Processes 437
16.3.2 Epidemic and Infrastructure 438
16.4 Recent Research Advances 439
16.4.1 Notation 439
16.4.2 Epidemic and Opinion Spreading Processes 440
16.4.2.1 Opinions Over Networks with Both Cooperative and Antagonistic Interactions 440
16.4.2.2 Coupled Epidemic and Opinion Dynamics 441
16.4.2.3 Opinion-Dependent Reproduction Number 443
16.4.2.4 Simulations 444
16.4.3 Epidemic Spreading with Shared Resources 445
16.4.3.1 The Multi-Virus SIWS Model 445
16.4.3.2 Problem Statements 447
16.4.3.3 Analysis of the Eradicated State of a Virus 448
16.4.3.4 Persistence of a Virus 449
16.4.3.5 Simulations 449
16.5 Future Research Challenges and Visions 450
References 451
17 Digital Twins and Automation of Care in the Intensive Care Unit 457
J. Geoffrey Chase, Cong Zhou, Jennifer L. Knopp, Knut Moeller, Balázs Benyo, Thomas Desaive, Jennifer H. K. Wong, Sanna Malinen, Katharina Naswall, Geoffrey M. Shaw, Bernard Lambermont, and Yeong S. Chiew
17.1 Introduction 457
17.1.1 Economic Context 458
17.1.2 Healthcare Context 459
17.1.3 Technology Context 460
17.1.4 Overall Problem and Need 460
17.2 Digital Twins and CPHS 461
17.2.1 Digital Twin/Virtual Patient Definition 461
17.2.2 Requirements in an ICU Context 463
17.2.3 Digital Twin Models in Key Areas of ICU Care and Relative to Requirements 464
17.2.4 Review of Digital Twins in Automation of ICU Care 466
17.2.5 Summary 467
17.3 Role of Social-Behavioral Sciences 467
17.3.1 Introduction 467
17.3.2 Barriers to Innovation Adoption 467
17.3.3 Ergonomics and Codesign 468
17.3.4 Summary (Key Takeaways) 469
17.4 Future Research Challenges and Visions 470
17.4.1 Technology Vision of the Future of CPHS in ICU Care 470
17.4.2 Social-Behavioral Sciences Vision of the Future of CPHS in ICU Care 471
17.4.3 Joint Vision of the Future and Challenges to Overcome 473
17.5 Conclusions 473
References 474
Part V Sociotechnical Systems 491
18 Online Attention Dynamics in Social Media 493
Maria Castaldo, Paolo Frasca, and Tommaso Venturini
18.1 Introduction to Attention Economy and Attention Dynamics 493
18.2 Online Attention Dynamics 494
18.2.1 Collective Attention Is Limited 494
18.2.2 Skewed Attention Distribution 495
18.2.3 The Role of Novelty 496
18.2.4 The Role of Popularity 496
18.2.5 Individual Activity Is Bursty 499
18.2.6 Recommendation Systems Are the Main Gateways for Information 500
18.2.7 Change Is the Only Constant 500
18.3 The New Challenge: Understanding Recommendation Systems Effect in Attention Dynamics 501
18.3.1 Model Description 502
18.3.2 Results and Discussion 503
18.4 Conclusion 505
Acknowledgments 505
References 505
19 Cyber–Physical–Social Systems for Smart City 511
Gang Xiong, Noreen Anwar, Peijun Ye, Xiaoyu Chen, Hongxia Zhao, Yisheng Lv, Fenghua Zhu, Hongxin Zhang, Xu Zhou, and Ryan W. Liu
19.1 Introduction 511
19.2 Social Community and Smart Cities 513
19.2.1 Smart Infrastructure 513
19.2.2 Smart Energy 515
19.2.3 Smart Transportation 515
19.2.4 Smart Healthcare 517
19.3 CPSS Concepts, Tools, and Techniques 518
19.3.1 CPSS Concepts 518
19.3.2 CPSS Tools 519
19.3.3 CPSS Techniques 520
19.3.3.1 IoT in Smart Cities 520
19.3.3.2 Big Data in Smart Cities 525
19.4 Recent Research Advances 528
19.4.1 Recent Research Advances of CASIA 528
19.4.2 Recent Research in European Union 531
19.4.3 Future Research Challenges and Visions 533
19.5 Conclusions 537
Acknowledgments 538
References 538
Part VI Concluding Remarks 543
20 Conclusion and Perspectives 545
Anuradha M. Annaswamy, Pramod P. Khargonekar, Françoise Lamnabhi-Lagarrigue, and Sarah K. Spurgeon
20.1 Benefits to Humankind: Synthesis of the Chapters and their Open Directions 545
20.2 Selected Areas for Current and Future Development in CPHS 547
20.2.1 Driver Modeling for the Design of Advanced Driver Assistance Systems 547
20.2.2 Cognitive Cyber–Physical Systems and CPHS 547
20.2.3 Emotion–Cognition Interactions 548
20.3 Ethical and Social Concerns: Few Directions 549
20.3.1 Frameworks for Ethics 550
20.3.2 Technical Approaches 550
20.4 Afterword 551
References 551
Index 555
