Decision-Making Techniques for Autonomous Vehicles (eBook)

Felipe Jimenez, Jorge Villagra (Herausgeber)

eBook Download: PDF | EPUB

2023 | 1. Auflage
424 Seiten
Elsevier Science (Verlag)
978-0-323-98549-9 (ISBN)

Infos

Decision-Making Techniques for Autonomous Vehicles provides a general overview of control and decision-making tools that could be used in autonomous vehicles. Motion prediction and planning tools are presented, along with the use of machine learning and adaptability to improve performance of algorithms in real scenarios. The book then examines how driver monitoring and behavior analysis are used produce comprehensive and predictable reactions in automated vehicles. The book ultimately covers regulatory and ethical issues to consider for implementing correct and robust decision-making. This book is for researchers as well as Masters and PhD students working with autonomous vehicles and decision algorithms. - Provides a complete overview of decision-making and control techniques for autonomous vehicles - Includes technical, physical, and mathematical explanations to provide knowledge for implementation of tools - Features machine learning to improve performance of decision-making algorithms - Shows how regulations and ethics influence the development and implementation of these algorithms in real scenarios

Inhalt

Front Cover 1
Decision-making Techniques for Autonomous Vehicles 4
Copyright 5
Contents 6
Contributors 12
About the editors 14
Chapter 1: Overview 18
1.1. Introduction 18
1.2. Decision-making, automation levels, and operational design domains 20
1.3. Scope of the book 21
1.4. Book structure overview 24
References 30
Part I: Embedded decision components 32
Chapter 2: Embodied decision architectures 34
2.1. Introduction 34
2.2. Embodiment and cognitive capabilities 34
2.3. Cognitive architectures and biological plausible human behavioral models 37
2.4. Decision architectures for autonomous driving 40
2.4.1. Examples of subsumption architectures 41
2.4.2. An ADAS-oriented behavior architecture 44
2.4.3. Examples of cognition-inspired architectures 45
2.4.4. Safety-oriented architectures 47
2.4.5. Shared control architectures 49
2.5. Common functional blocks 51
References 53
Chapter 3: Behavior planning 56
3.1. Introduction 56
3.2. Problem description 57
3.3. Automata and Markov processes 60
3.4. Basic decision theory 61
3.5. Sequential decision-making 65
3.6. Applications in automated vehicles 67
3.6.1. Rule-based planning 68
3.6.2. Reactive planning 69
3.6.3. Interaction-aware planning 70
3.6.4. Game theory for behavior planning 71
3.6.5. AI-enabled behavior planning 72
References 74
Chapter 4: Motion prediction and risk assessment 78
4.1. Introduction 78
4.1.1. Problem formalization 79
4.2. Driver trait estimation 81
4.2.1. Scope 81
4.2.2. Representation 81
4.2.3. Inference approaches 82
4.3. Intention estimation 82
4.3.1. Scope 83
4.3.2. Representation 84
4.3.3. Inference approaches 84
4.3.3.1. Recursive estimation algorithms 84
4.3.3.2. Bayesian approaches 84
4.3.3.3. Game-theoretic approaches 85
4.3.3.4. Learning-based approaches 85
4.4. Motion prediction 91
4.4.1. Scope 91
4.4.2. Representation 92
4.4.2.1. Agent-level representation 94
4.4.2.2. Scenario level representation 95
4.4.3. Modeling approaches 96
4.4.3.1. Physics-based 96
Single-model approaches 96
Multimodel approaches 98
4.4.3.2. Learning-based approaches 99
Sequential 99
Nonsequential 101
4.4.3.3. Planning-based 102
Forward planning 102
Inverse planning 102
4.4.4. Situational awareness considerations 103
4.4.4.1. Unaware 103
4.4.4.2. Intention-aware 103
4.4.4.3. Scene awareness 104
4.4.4.4. Map aware 104
4.4.5. Metrics 105
4.4.5.1. Geometric accuracy metrics 105
4.4.5.2. Probabilistic accuracy metrics 106
4.5. Risk assessment 106
4.5.1. Scope 106
4.5.2. Representation 107
4.5.3. Inference strategies 108
4.5.3.1. Risk based on future trajectories 108
Binary collision prediction 108
Collision risk based on indicators 109
Probabilistic collision prediction 110
4.5.4. Risk based on unexpected behavior 111
4.5.4.1. Detecting unusual events 111
4.5.4.2. Detecting conflicting maneuvers 112
References 112
Chapter 5: Motion search space 120
5.1. Introduction 120
5.2. Graph-based techniques 121
5.3. Geometric methods 122
5.3.1. Nonobstacle-based techniques 124
5.3.2. Obstacle-based techniques 125
5.4. Sampling-based methods 126
5.5. Driving corridors 130
References 131
Chapter 6: Motion planning 134
6.1. Problem definition 134
6.1.1. Brief taxonomy of motion planners 136
6.2. Geometric methods 138
6.2.1. Point-free template-based geometric strategies 138
6.2.2. Point-based template-based curves 140
6.3. Variational and optimal methods 142
6.3.1. MPC architecture 144
6.3.2. Optimization techniques 145
6.3.3. Local nonconvex optimization 146
6.3.4. Global nonconvex optimization 147
6.3.5. Relevant uses cases 152
6.3.5.1. Strategies considering comfort and safety 152
6.3.5.2. Obstacle avoidance and overtaking maneuvers 153
6.4. Sampling-based methods 155
6.4.1. General formulation of the deterministic problem 155
6.4.1.1. Metric state and measure function 156
6.4.1.2. Sampling strategy 156
6.4.1.3. Collision detection and path segment validation 157
6.4.2. Multiple-query and single-query methods 158
6.4.2.1. Single-query methods 158
6.4.2.2. Multiple-query methods 158
6.4.2.3. Deterministic approaches in autonomous driving 160
6.4.3. General formulation of the probabilistic problem 161
6.4.3.1. Probabilistic approaches in autonomous driving 161
6.4.4. Sampling-based methods with constraints 162
6.5. Graph-search methods 163
6.6. Cognition-inspired approaches 165
6.6.1. Evolutionary computation 166
6.6.2. Fuzzy logic and neural networks 167
6.7. Biomimetic methods 167
6.7.1. Artificial potential fields 167
6.7.2. Elastic bands 171
6.8. From vehicle following/CACC to standalone speed planning 172
6.9. Separated speed planning 174
6.10. Joint path and speed optimization-based planning 177
References 178
Chapter 7: End-to-end architectures 186
7.1. End-to-end approaches 186
7.1.1. Introduction 186
7.2. End-to-end approaches based on deep learning 188
7.2.1. Classification of end-to-end architectures for autonomous driving 190
7.2.1.1. SiD-E2E architecture 190
7.2.1.2. MiD-E2E architecture 191
7.2.1.3. SeD-E2E architecture 192
7.2.2. Transfer learning vs ad-hoc solutions 194
7.2.2.1. Transfer learning 194
7.2.2.2. Ad-hoc solution 195
7.2.3. Datasets for modeling end-to-end solutions 196
7.2.4. Reinforcement learning techniques 199
7.3. Expert systems 203
7.4. Future perspectives 205
References 206
Chapter 8: Interplay between decision and control 210
8.1. Introduction 210
8.2. Stabilization principles 211
8.2.1. Problem definition 211
8.2.2. Stabilization requirements for automated driving 212
8.2.3. Longitudinal control 214
8.2.4. Lateral control 216
8.3. Upstream vs downstream control architectures 218
8.4. Interaction models between planning and control 222
8.4.1. Ethical considerations 222
8.4.2. Integrated vs decoupled planning and control 224
8.5. Fail-operational considerations 226
References 227
Part II: Infrastructure-oriented decision-making 232
Chapter 9: Traffic data analysis and route planning 234
9.1. Introduction 234
9.2. Off-board decision-making: From the traveling salesman problem to the vehicle routing problem 235
9.3. On the relevance of traffic data and exogenous information for predictive route planning 242
9.3.1. Considering traffic forecasts 243
9.3.2. Short-term traffic forecasting 243
9.3.3. Long-term traffic forecasting 248
9.4. Challenges and research directions in the confluence between route planning and traffic data analysis 250
9.4.1. From route optimization toward learning to route 250
9.4.2. Causal agent-based traffic models and route planning 252
9.4.3. Knowledge transfer for route optimization 253
9.4.4. Toward explainable and trustworthy route planning 254
References 255
Chapter 10: Cooperative driving 262
10.1. Introduction to cooperative, connected, and automated driving (3P) 262
10.1.1. Solution: CCAD 264
10.1.1.1. Onboard decision-making 264
10.1.1.2. Decision-making in infrastructure 265
10.2. Communication technologies 265
10.2.1. Vehicle-to-everything (V2X) 266
10.2.2. DSRC (IEEE 802.11p, ETSI ITS-G5) (V2X standards overview) 267
10.2.2.1. IEEE 802.11p 268
10.2.3. Cellular V2X 268
10.2.4. Security 269
10.2.4.1. SerIoT project 270
10.3. Connected services 271
10.4. Adaptation of decision-making mechanisms to support V2X 273
10.4.1. Connected and automated scenario for cyber-attacks (SerIoT project) 273
10.4.1.1. Smart intersection in normal situation 274
10.4.1.2. Fleet management in normal situation 275
10.4.2. Platoon maneuver 276
10.4.3. Roundabout merging scenarios maneuvers 277
10.4.4. Conclusions 277
References 278
Chapter 11: Infrastructure impact 280
11.1. The role of the physical infrastructure: From evidence to guidelines 280
11.1.1. Impact of road infrastructure on automated driving 280
11.1.1.1. Road typology 280
11.1.1.2. Geometry 281
11.1.1.3. Road markings 282
11.1.1.4. Traffic signs 284
11.1.1.5. Junctions 284
11.1.1.6. Pavement condition 285
11.1.1.7. Road environment 285
11.1.1.8. Environmental conditions 286
11.1.1.9. Road works and temporary emergency signage 286
11.1.1.10. Speed 286
11.2. Information required from infrastructure to enable different ad levels 287
11.2.1. Specifications of road infrastructure for different AD levels 287
11.2.1.1. Level of service for automated driving (LOSAD) 287
11.2.1.2. Infrastructure support levels for automated driving (ISAD) 289
11.2.1.3. Smart road levels (SRL) 289
11.2.1.4. Factors related to physical infrastructure 291
11.2.1.5. Information management providers 292
11.2.2. Minimal risk condition 293
References 294
Part III: User influence 298
Chapter 12: Driver behavior 300
12.1. A human-centered perspective in driving automation 300
12.2. Human-driver assessment from the perspective of HAI models in automated driving 310
12.2.1. Definition and assessment of mental workload (MWL) 311
12.2.2. Reduced situation awareness (SA) 314
12.2.3. Complacency or overtrust 316
12.2.4. Skill degradation and the loss of sense of authority 317
12.2.5. Control transition between automation and human driver in automated vehicles 317
12.3. Passengers in autonomous vehicles 322
12.3.1. The changing role of the passenger in autonomous vehicles 322
12.3.2. The acceptance of autonomous vehicles 323
12.3.2.1. Personal expectations 324
12.3.2.2. Perceived attributes 325
12.3.2.3. Personality factors 325
12.3.3. Emotional state of the passenger in autonomous vehicles 326
12.3.4. Driver attributes and external factors that affect the passenger state 334
12.3.4.1. External factors 334
12.3.4.2. Ego vehicle factors 335
12.3.4.3. Internal factors 335
12.3.5. Human machine interface for passengers in autonomous vehicles 336
12.3.6. Ride, ambient comfort, well-being, and other services 337
References 341
Chapter 13: Human-machine interaction 350
13.1. Introduction 350
13.2. Human-Machine Cooperation and metaphors used in shared control 352
13.3. Shared control approaches 353
13.3.1. Definitions 353
13.3.2. Frameworks 355
13.3.3. Algorithms 356
13.4. A recent human-machine interaction framework 357
13.5. Traded control mechanisms 360
13.5.1. Suitability of traded control 360
13.5.2. Traded control (de)activation principles 363
References 366
Part IV: Deployment issues 370
Chapter 14: Algorithms validation 372
14.1. Introduction 372
14.2. Validation methodology 373
14.2.1. Testing process 374
14.2.2. Main techniques for ADF validation 375
14.2.3. Datasets 377
14.3. Simulation systems 378
14.3.1. Human-in-the-Loop simulation (HITL) 380
14.3.1.1. Human-in-the-Loop AI 381
14.3.1.2. Human-in-the-Loop driving simulation 382
14.3.2. Vehicle-in-the-loop simulation 383
14.3.2.1. Definition 383
14.3.2.2. Traffic simulation: Scene simulation 385
14.4. Standards for safety assurance 386
References 387
Chapter 15: Legal and social aspects 392
15.1. Introduction 392
15.2. Regulation 392
15.2.1. Introduction 393
15.2.2. International governance 394
15.2.3. The Vienna convention on road traffic 395
15.2.3.1. Amendments to the Vienna convention on road traffic 396
15.2.4. State-of-the-art on the European regulations on autonomous vehicles 397
15.2.4.1. Context of European regulations 397
15.2.4.2. Comparative analysis of international regulations 398
15.3. Ethics 398
15.3.1. Ethical problem of autonomous driving 398
15.3.2. Approaches to face the ethical problem 402
15.3.3. Conclusions 407
15.4. User acceptance 407
15.4.1. Introduction 407
15.4.2. Perceived safety 409
15.4.3. Trust 410
15.4.4. Demographic factors 411
15.4.5. Psychological factors 412
References 413
Index 418
Back Cover 426

Produkt-Details

• Erscheint lt. Verlag: 3.3.2023
• Sprache: englisch
• Themenwelt: Technik ► Fahrzeugbau / Schiffbau
• ISBN-10: 0-323-98549-1 / 0323985491
• ISBN-13: 978-0-323-98549-9 / 9780323985499