本书首先介绍人工智能机器人的定义、历史和体系结构，然后全面系统地阐述人工智能机器人在传感、感知、运动、规划、导航、学习、交互等方面的基础理论和关键技术。全书共分为五部分。**部分共5章，定义了什么是智能机器人，介绍了人工智能机器人简史，并讨论了自动化与自治、软件体系结构和遥操作；*二部分共6章，针对机器人的反应（行为）层智能展开讨论，分别对应机器人行为、感知与行为、行为协调、运动学、传感器与感知，以及距离感知等方面的内容；第三部分共5章，详细讨论机器人的慎思层智能，包括慎思层的内涵、导航、路径和动作规划、定位、建图与探索，以及机器学习等内容；第四部分共2章，讨论机器人的交互层智能，包括多机器人系统和人-机器人交互；第五部分共2章，分别介绍自治系统的设计与评估方法，以及与机器人相关的伦理问题。

I Framework for Thinking About AI and Robotics 
1 What Are Intelligent Robots? 
1.1 Overview 
1.2 Definition: What Is an Intelligent Robot? 
1.3 What Are the Components of a Robot? 
1.4 Three Modalities: What Are the Kinds of Robots? 
1.5 Motivation: Why Robots? 
1.6 Seven Areas of AI: Why Intelligence? 
1.7 Summary 
1.8 Exercises 
1.9 End Notes 
2 A Brief History of AI Robotics 
2.1 Overview 
2.2 Robots as Tools, Agents, or Joint Cognitive Systems 
2.3 World War II and the Nuclear Industry 
2.4 Industrial Manipulators 
2.5 Mobile Robots 
2.6 Drones 
2.7 The Move to Joint Cognitive Systems 
2.8 Summary 
2.9 Exercises 
2.10 End Notes 
3 Automation and Autonomy 
3.1 Overview 
3.2 The Four Sliders of Autonomous Capabilities 
3.2.1 Plans: Generation versus Execution 
3.2.2 Actions: Deterministic versus Non-deterministic 
3.2.3 Models: Open- versus Closed-World 
3.2.4 Knowledge Representation: Symbols versus Signals 
3.3 Bounded Rationality 
3.4 Impact of Automation and Autonomy 
3.5 Impact on Programming Style 
3.6 Impact on Hardware Design 
3.7 Impact on Types of Functional Failures 
3.7.1 Functional Failures 
3.7.2 Impact on Types of Human Error 
3.8 Trade-Spaces in Adding Autonomous Capabilities 
3.9 Summary 
3.10 Exercises 
3.11 End Notes 
4 Software Organization of Autonomy 
4.1 Overview 
4.2 The Three Types of Software Architectures 
4.2.1 Types of Architectures 
4.2.2 Architectures Reinforce Good Software Engineering Principles 
4.3 Canonical AI Robotics Operational Architecture 
4.3.1 Attributes for Describing Layers 
4.3.2 The Reactive Layer 
4.3.3 The Deliberative Layer 
4.3.4 The Interactive Layer 
4.3.5 Canonical Operational Architecture Diagram 
4.4 Other Operational Architectures 
4.4.1 Levels of Automation 
4.4.2 Autonomous Control Levels (ACL) 
4.4.3 Levels of Initiative 
4.5 Five Subsystems in Systems Architectures 
4.6 Three Systems Architecture Paradigms 
4.6.1 Trait 1: Interaction Between Primitives 
4.6.2 Trait 2: Sensing Route 
4.6.3 Hierarchical Systems Architecture Paradigm 
4.6.4 Reactive Systems Paradigm 
4.6.5 Hybrid Deliberative/Reactive Systems Paradigm 
4.7 Execution Approval and Task Execution 
4.8 Summary 
4.9 Exercises 
4.10 End Notes 
5 Telesystems 
5.1 Overview 
5.2 Taskable Agency versus Remote Presence 
5.3 The Seven Components of a Telesystem 
5.4 Human Supervisory Control 
5.4.1 Types of Supervisory Control 
5.4.2 Human Supervisory Control for Telesystems 
5.4.3 Manual Control 
5.4.4 Traded Control 
5.4.5 Shared Control 
5.4.6 Guarded Motion 
5.5 Human Factors 
5.5.1 Cognitive Fatigue 
5.5.2 Latency 
5.5.3 Human: Robot Ratio 
5.5.4 Human Out-of-the-Loop Control Problem 
5.6 Guidelines for Determining if a Telesystem Is Suitable for an Application 
5.6.1 Examples of Telesystems 
5.7 Summary 
5.8 Exercises 
5.9 End Notes 
II Reactive Functionality 
6 Behaviors 
6.1 Overview 
6.2 Motivation for Exploring Animal Behaviors 
6.3 Agency and Marr’s Computational Theory 
6.4 Example of Computational Theory: Rana Computatrix 
6.5 Animal Behaviors 
6.5.1 Reflexive Behaviors 
6.6 Schema Theory 
6.6.1 Schemas as Objects 
6.6.2 Behaviors and Schema Theory 
6.6.3 S-R: Schema Notation 
6.7 Summary 
6.8 Exercises 
6.9 End Notes 
7 Perception and Behaviors 
7.1 Overview 
7.2 Action-Perception Cycle 
7.3 Gibson: Ecological Approach 
7.3.1 Optic Flow 
7.3.2 Nonvisual Affordances 
7.4 Two Perceptual Systems 
7.5 Innate Releasing Mechanisms 
7.5.1 Definition of Innate Releasing Mechanisms 
7.5.2 Concurrent Behaviors 
7.6 Two Functions of Perception 
7.7 Example: Cockroach Hiding 
7.7.1 Decomposition 
7.7.2 Identifying Releasers 
7.7.3 Implicit versus Explicit Sequencing 
7.7.4 Perception 
7.7.5 Architectural Considerations 
7.8 Summary 
7.9 Exercises 
7.10 End Notes 
8 Behavioral Coordination 
8.1 Overview 
8.2 Coordination Function 
8.3 Cooperating Methods: Potential Fields 
8.3.1 Visualizing Potential Fields 
8.3.2 Magnitude Profiles 
8.3.3 Potential Fields and Perception 
8.3.4 Programming a Single Potential Field 
8.3.5 Combination of Fields and Behaviors 
8.3.6 Example Using One Behavior per Sensor 
8.3.7 Advantages and Disadvantages 
8.4 Competing Methods: Subsumption 
8.4.1 Example 
8.5 Sequences: Finite State Automata 
8.5.1 A Follow the Road FSA 
8.5.2 A Pick Up the Trash FSA 
8.6 Sequences: Scripts 
8.7 AI and Behavior Coordination 
8.8 Summary 
8.9 Exercises 
8.10 End Notes 
9 Locomotion 
9.1 Overview 
9.2 Mechanical Locomotion 
9.2.1 Holonomic versus Nonholonomic 
9.2.2 Steering 
9.3 Biomimetic Locomotion 
9.4 Legged Locomotion 
9.4.1 Number of Leg Events 
9.4.2 Balance 
9.4.3 Gaits 
9.4.4 Legs with Joints 
9.5 Action Selection 
9.6 Summary 
9.7 Exercises 
9.8 End Notes 
10 Sensors and Sensing 
10.1 Overview 
10.2 Sensor and Sensing Model 
10.2.1 Sensors: Active or Passive 
10.2.2 Sensors: Types of Output and Usage 
10.3 Odometry, Inertial Navigation System (INS) and Global Positioning System (GPS) 
10.4 Proximity Sensors 
10.5 Computer Vision 
10.5.1 Computer Vision Definition 
10.5.2 Grayscale and Color Representation 
10.5.3 Region Segmentation 
10.5.4 Color Histogramming 
10.6 Choosing Sensors and Sensing 
10.6.1 Logical Sensors 
10.6.2 Behavioral Sensor Fusion 
10.6.3 Designing a Sensor Suite 
10.7 Summary 
10.8 Exercises 
10.9 End Notes 
11 Range Sensing 
11.1 Overview 
11.2 Stereo 
11.3 Depth from X 
11.4 Sonar or Ultrasonics 
11.4.1 Light Stripers 
11.4.2 Lidar 
11.4.3 RGB-D Cameras 
11.4.4 Point Clouds 
11.5 Case Study: Hors d’Oeuvres, Anyone? 
11.6 Summary 
11.7 Exercises 
11.8 End Notes 
III Deliberative Functionality 
12 Deliberation 
12.1 Overview 
12.2 Strips 
12.2.1 More Realistic Strips Example 
12.2.2 Strips Summary 
12.2.3 Revisiting the Closed-World Assumption and the Frame Problem 
12.3 Symbol Grounding Problem 
12.4 GlobalWorld Models 
12.4.1 Local Perceptual Spaces 
12.4.2 Multi-level or HierarchicalWorld Models 
12.4.3 Virtual Sensors 
12.4.4 Global World Model and Deliberation 
12.5 Nested Hierarchical Controller 
12.6 RAPS and 3T 
12.7 Fault Detection Identification and Recovery 
12.8 Programming Considerations 
12.9 Summary 
12.10 Exercises 
12.11 End Notes 
13 Navigation 
13.1 Overview 
13.2 The Four Questions of Navigation 
13.3 Spatial Memory 
13.4 Types of Path Planning 
13.5 Landmarks and Gateways 
13.6 Relational Methods 
13.6.1 Distinctive Places 
13.6.2 Advantages and Disadvantages 
13.7 Associative Methods 
13.8 Case Study of Topological Navigation with a Hybrid Architecture 
13.8.1 Topological Path Planning 
13.8.2 Navigation Scripts 
13.8.3 Lessons Learned 
13.9 Discussion of Opportunities for AI 
13.10 Summary 
13.11 Exercises 
13.12 End Notes 
14 Metric Path Planning and Motion Planning 
14.1 Overview 
14.2 Four Situations Where Topological Navigation Is Not Sufficient 
14.3 Configuration Space 
14.3.1 Meadow Maps 
14.3.2 Generalized Voronoi Graphs 
14.3.3 Regular Grids 
14.3.4 Quadtrees 
14.4 Metric Path Planning 
14.4.1 A* and Graph-Based Planners 
14.4.2 Wavefront-Based Planners 
14.5 Executing a Planned Path 
14.5.1 Subgoal Obsession 
14.5.2 Replanning 
14.6 Motion Planning 
14.7 Criteria for Evaluating Path and Motion Planners 
14.8 Summary 
14.9 Exercises 
14.10 End Notes 
15 Localization, Mapping, and Exploration 
15.1 Overview 
15.2 Localization 
15.3 Feature-Based Localization 
15.4 Iconic Localization 
15.5 Static versus Dynamic Environments 
15.6 Simultaneous Localization and Mapping 
15.7 Terrain Identification and Mapping 
15.7.1 Digital Terrain Elevation Maps 
15.7.2 Terrain Identification 
15.7.3 Stereophotogrammetry 
15.8 Scale and Traversability 
15.8.1 Scale 
15.8.2 Traversability Attributes 
15.9 Exploration 
15.9.1 Reactive Exploration 
15.9.2 Frontier-Based Exploration 
15.9.3 Generalized Voronoi Graph Methods 
15.10 Localization, Mapping, Exploration, and AI 
15.11 Summary 
15.12 Exercises 
15.13 End Notes 
16 Learning 
16.1 Overview 
16.2 Learning 
16.3 Types of Learning by Example 
16.4 Common Supervised Learning Algorithms 
16.4.1 Induction 
16.4.2 Support Vector Machines 
16.4.3 Decision Trees 
16.5 Common Unsupervised Learning Algorithms 
16.5.1 Clustering 
16.5.2 Artificial Neural Networks 
16.6 Reinforcement Learning 
16.6.1 Utility Functions 
16.6.2 Q-learning 
16.6.3 Q-learning Example 
16.6.4 Q-learning Discussion 
16.7 Evolutionary Robotics and Genetic Algorithms 
16.8 Learning and Architecture 
16.9 Gaps and Opportunities 
16.10 Summary 
16.11 Exercises 
16.12 End Notes 
IV Interactive Functionality 
17 MultiRobot Systems (MRS) 
17.1 Overview 
17.2 Four Opportunities and Seven Challenges 
17.2.1 Four Advantages of MRS 
17.2.2 Seven Challenges in MRS 
17.3 Multirobot Systems and AI 
17.4 Designing MRS for Tasks 
17.4.1 Time Expectations for a Task 
17.4.2 Subject of Action 
17.4.3 Movement 
17.4.4 Dependency 
17.5 Coordination Dimension of MRS Design 
17.6 Systems Dimensions in Design 
17.6.1 Communication 
17.6.2 MRS Composition 
17.6.3 Team Size 
17.7 Five Most Common Occurrences of MRS 
17.8 Operational Architectures for MRS 
17.9 Task Allocation 
17.10 Summary 
17.11 Exercises 
17.12 End Notes 
18 Human-Robot Interaction 
18.1 Overview 
18.2 Taxonomy of Interaction 
18.3 Contributions from HCI, Psychology, Communications 
18.3.1 Human-Computer Interaction 
18.3.2 Psychology 
18.3.3 Communications 
18.4 User Interfaces 
18.4.1 Eight Golden Rules for User Interface Design 
18.4.2 Situation Awareness 
18.4.3 Multiple Users 
18.5 Modeling Domains, Users, and Interactions 
18.5.1 Motivating Example of Users and Interactions 
18.5.2 Cognitive Task Analysis 
18.5.3 CognitiveWork Analysis 
18.6 Natural Language and Naturalistic User Interfaces 
18.6.1 Natural Language Understanding 
18.6.2 Semantics and Communication 
18.6.3 Models of the Inner State of the Agent 
18.6.4 Multi-modal Communication 
18.7 Human-Robot Ratio 
18.8 Trust 
18.9 Testing and Metrics 
18.9.1 Data Collection Methods 
18.9.2 Metrics 
18.10 Human-Robot Interaction and the Seven Areas of Artificial Intelligence 
18.11 Summary 
18.12 Exercises 
18.13 End Notes 
V Design and the Ethics of Building Intelligent Robots 
19 Designing and Evaluating Autonomous Systems 
19.1 Overview 
19.2 Designing a Specific Autonomous Capability 
19.2.1 Design Philosophy 
19.2.2 Five Questions for Designing an Autonomous Robot 
19.3 Case Study: Unmanned Ground Robotics Competition 
19.4 Taxonomies and Metrics versus System Design 
19.5 Holistic Evaluation of an Intelligent Robot 
19.5.1 Failure Taxonomy 
19.5.2 Four Types of Experiments 
19.5.3 Data to Collect 
19.6 Case Study: Concept Experimentation 
19.7 Summary 
19.8 Exercises 
19.9 End Notes 
20 Ethics 
20.1 Overview 
20.2 Types of Ethics 
20.3 Categorizations of Ethical Agents 
20.3.1 Moor’s Four Categories 
20.3.2 Categories of Morality 
20.4 Programming Ethics 
20.4.1 Approaches from Philosophy 
20.4.2 Approaches from Robotics 
20.5 Asimov’s Three Laws of Robotics 
20.5.1 Problems with the Three Laws 
20.5.2 The Three Laws of Responsible Robotics 
20.6 Artificial Intelligence and Implementing Ethics 
20.7 Summary 
20.8 Exercises 
20.9 End Notes 
Bibliography 
Index

分别于1980年、1989年和1992年在美国佐治亚理工学院获得机械工程学学士学位、计算机科学硕士和博士学位，现任德克萨斯农工大学计算机科学与工程系的Raytheon荣誉教授，机器人辅助搜索与救援研究中心主任，IEEE会士，曾任IEEE机器人和自动化执行委员会执委。研究方向为人工智能，人-机器人交互，以及异构多机器人系统。已发表100多部/篇出版物，是国际上救援机器人和人-机器人交互领域的开创者之一。 

Robin R. Murphy分别于1980年、1989年和1992年在美国佐治亚理工学院获得机械工程学学士学位、计算机科学硕士和博士学位，现任得克萨斯农工大学计算机科学与工程系Raytheon荣誉教授，机器人辅助搜索与救援研究中心主任。IEEE会士，曾任IEEE机器人和自动化执行委员会执委。研究方向为人工智能、人-机器人交互，以及异构多机器人系统。已发表100多部/篇出版物，是国际范围内救援机器人和人-机器人交互领域的开创者之一。

本书可作为机器人工程、人工智能、电子、自动化及计算机等专业高年级本科生和研究生的教材或参考书，也可供从事智能机器人方面研究的教师和研究人员，以及参加各类机器人学科竞赛的指导教师和学生学习参考。