Knowledge Engineering Tools and Techniques for AI Planning (eBook)

Mauro Vallati, Diane Kitchin (Herausgeber)

eBook Download: PDF

2020 | 1st ed. 2020
VIII, 277 Seiten
Springer International Publishing (Verlag)
978-3-030-38561-3 (ISBN)

Infos

This book presents a comprehensive review for Knowledge Engineering tools and techniques that can be used in Artificial Intelligence Planning and Scheduling. KE tools can be used to aid in the acquisition of knowledge and in the construction of domain models, which this book will illustrate. 

AI planning engines require a domain model which captures knowledge about how a particular domain works - e.g. the objects it contains and the available actions that can be used. However, encoding a planning domain model is not a straightforward task - a domain expert may be needed for their insight into the domain but this information must then be encoded in a suitable representation language. The development of such domain models is both time-consuming and error-prone. Due to these challenges, researchers have developed a number of automated tools and techniques to aid in the capture and representation of knowledge.

This book targets researchers and professionals working in knowledge engineering, artificial intelligence and software engineering. Advanced-level students studying AI will also be interested in this book.

Dr Diane Kitchin has worked as a Senior Lecturer in the Department of Computer Science at the University of Huddersfield since 2000.  Her main research focuses on the area of AI Planning and Knowledge Engineering.  She has published papers on Object-Centred Planning, Tools for AI, Portfolio-based planning and Domain model acquisition in a number of conference proceedings.  Work on Planning Domain Definition appeared in the Journal of Knowledge Engineering, with further journal publications in the International Journal on Artificial Intelligence Tools, The Knowledge Engineering Review and AI Communications.

Dr Mauro Vallati is a Senior Lecturer in the Department of Computer Science at the University of Huddersfield. He has extensive experience in real-world applications of AI methods and techniques, with his research focusing on the Knowledge Engineering aspects of AI applications. Among the others, he investigated the use of AI for managing urban traffic control, for controlling robots, and for reducing the energy consumption of manufacturing machine tools. Dr. Vallati has published a significant number of papers in top AI venues, and has co-organised important events for the AI field, such as workshops, competitions, and conferences. He delivered numerous tutorials in important AI venues.

Inhalt

Preface 5
Contents 7
Part I Knowledge Capture and Encoding 9
1 Explanation-Based Learning of Action Models 10
1 Introduction 10
2 Background 11
2.1 Classical Planning with Conditional Effects 11
2.2 The Observation Model 12
2.3 Explaining Observations with Classical Planning 13
3 Explanation-Based Learning of Strips Action Models 14
3.1 The Space of Strips Action Models 14
3.2 The Sampling Space 16
4 Learning Strips Action Models with Classical Planning 17
4.1 Compilation 17
4.2 Properties of the Compilation 22
5 Experimental Results 23
5.1 Learning from Labeled Plans 24
5.2 Learning from Initial/Final State Pairs 24
6 Conclusions 25
References 26
2 Automated Domain Model Learning Tools for Planning 28
1 Introduction 28
1.1 Knowledge Representation for Knowledge Engineering of Domain Models 30
2 Domain Model Learning Techniques and Tools 31
2.1 Inductive Learning 32
2.1.1 When to Use Inductive Learning 34
2.2 Knowledge-Based Inductive Learning (KBIL) 35
2.3 Analytical Learning 38
2.4 Hybrid Learning 39
2.5 Surprise-Based Learning (SBL) 40
2.6 Transfer Learning 42
2.7 Policy Learning 43
2.8 Other Methods of Knowledge Acquisition 44
3 Characteristics of the Domain Model Learning Tools 46
4 Conclusion 49
References 50
3 Formal Knowledge Engineering for Planning: Pre and Post-Design Analysis 54
1 Introduction 55
2 Knowledge Engineering and Planning 56
3 Domain Modeling in AI Planning 58
3.1 Accuracy 59
3.2 Adequacy 60
3.3 Operationality 61
4 A Knowledge Engineering Design Approach for Planning 63
5 PDM and Post-Design Modeling Using Petri Nets 66
6 New Perspectives for AI Planning in Automation Systems 69
References 70
4 MyPDDL: Tools for Efficiently Creating PDDL Domains and Problems 73
1 Introduction 73
2 Related Work 75
2.1 Critical Review 78
3 MyPDDL 80
3.1 Modules 80
4 Validation and Evaluation 86
4.1 User Evaluation 86
4.1.1 Analysis 88
4.1.2 Results 88
5 Conclusion 91
Appendix: Tasks 92
Deliberately Erroneous Logistics Domain 92
Deliberately Erroneous Coffee Domain 93
Planet Splisus 94
Store 95
References 95
5 KEPS Book: Planning.Domains 97
1 Planning.Domains Solver 98
1.1 Libraries 99
1.2 API Future 101
2 Solver Planning Domains 102
2.1 Solver Future 103
3 Editor Planning Domains 103
3.1 Plugin Framework 104
3.2 Session Functionality 105
3.3 Editor Future 105
4 Education Planning Domains 106
5 What Is Next for Planning.Domains 106
5.1 Planimation 106
5.2 VSCode Integration 109
6 Conclusion 110
References 110
6 Modeling Planning Tasks: Representation Matters 112
1 Introduction 112
2 Outer Entanglements 113
3 Macro-Operators 115
4 Bagged Representation 118
5 Procedural Domain Control Knowledge 120
6 Transition-Based Domain Control Knowledge 121
7 A Case Study: The Spanner Domain 124
8 Conclusion 125
References 126
Part II Interaction, Visualisation, and Explanation 129
7 An Interactive Tool for Plan Generation, Inspection, and Visualization 130
1 Introduction 130
2 Preliminaries 133
2.1 The Planning Problem 133
2.2 The LPG Planner 136
2.2.1 Plan Representation Through LA-Graphs 137
2.2.2 Local Search in the Space of LA-Graphs 138
3 Architecture of InLPG 139
3.1 Architecture Overview 139
3.2 Input Module 142
3.3 Search Process Monitor 142
3.4 Search State Monitor 143
3.5 Plan Editor 145
3.6 Search Process Editor 146
4 Walk-through Example of a User Interaction 147
5 Experiments 152
6 Related Work 154
7 Conclusions 155
References 156
8 Interactive Visualization in Planning and Scheduling 159
1 Introduction 159
2 Interactive Gantt Chart (iGantt) 160
2.1 Problem Specification 160
2.2 Visualization of Schedules 161
2.3 Interactive Schedule Modifications 162
2.4 Automated Schedule Repair 162
3 Interactive Workflow Optimization (FlowOpt) 165
4 Interactive Visualization and Verification of Plan (VisPlan) 167
4.1 Plan Verification 168
4.2 Visualization of Sequential and Temporal Plans 169
4.2.1 Visualization of STRIPS Plans 171
4.2.2 Visualization of Temporal Plans 171
4.3 Interactive Plan Modifications 172
5 Conclusions 173
References 173
9 Argument-Based Plan Explanation 175
1 Introduction 175
2 Argumentation and Dialogue 176
2.1 Abstract Argumentation 177
2.2 Labellings 178
2.3 From Knowledge to Arguments 178
3 Proof Dialogues 181
4 Putting it all Together: The SAsSy Demonstrator 184
4.1 Plan Visualisation 185
4.2 Natural Language Generation 185
4.3 Dialogue Based Plan Explanation 186
5 Discussion and Related Work 187
6 Conclusions 188
References 188
10 Interactive Planning-Based Hypothesis Generation with LTS++ 191
1 Introduction and Motivation 192
2 Application Description 193
3 Hypothesis Generation Problem 196
4 Model Description in LTS++ 199
4.1 From LTS++ to a Planning Problem in PDDL 201
5 LTS++ Integrated Development Environment 204
6 Related Work 207
7 Summary 207
References 208
11 Web Planner: A Tool to Develop, Visualize, and Test Classical Planning Domains 210
1 Introduction 210
2 Background 212
2.1 Planning 212
2.2 Data Visualization 212
3 Web Planner Architecture 213
3.1 Domain Development Interface 214
3.2 Visualization Interface 216
4 Deployment and Evaluation 220
4.1 Case Study 220
4.2 Case Study Survey Results 222
4.3 General Public Usage Statistics 223
5 Related Work 223
6 Conclusions 226
References 227
Part III Case Studies and Applications 229
12 Design of Timeline-Based Planning Systems for Safe Human-Robot Collaboration 230
1 Introduction 230
2 Fostering Autonomy via Timeline-Based Planning and Execution 232
2.1 A Theoretical Framework 232
2.2 PLATINUm: A Timeline-Based Planning and Acting Framework 234
3 KeeN: Knowledge Engineering ENvironment 235
3.1 Knowledge Engineering and Verification and Validation Features in KeeN 236
4 Deploying Task Planning Solutions for Safe Human-Robot Collaboration 238
4.1 A Specific Human-Robot Collaboration Case Study 239
4.2 An Engineering and Control Architecture for HRC 240
4.3 The FourByThree Controller 242
4.4 Implementation with a Real Robot 244
5 Conclusions 245
References 246
13 Planning in a Real-World Application: An AUV Case Study 248
1 Introduction 248
2 Background 249
3 One-Shot Planning 250
3.1 Requirements 250
3.2 Domain Model Specification 251
3.3 Problem Specification 252
3.4 Field Experiment 253
4 Dynamic Planning, Replanning, and Plan Execution 253
4.1 Requirements 254
4.2 Domain Model Specification 254
4.3 Problem Specification 256
4.4 Planning and Execution 256
4.5 Field Experiment 256
5 Conclusion 257
References 257
14 Knowledge Engineering and Planning for Social Human–Robot Interaction: A Case Study 259
1 Introduction 259
2 Interaction Management 261
3 Task-Based Social Interaction: A Robot Bartender Scenario 262
4 Modelling Social Human–Robot Interaction for Planning 263
4.1 Planning with Knowledge and Sensing 263
4.2 State Management 265
4.3 Representing Properties, Actions, Objects, and Goals 266
5 Planning for Social Human–Robot Interaction 268
5.1 Ordering a Drink 268
5.2 Ordering Drinks with Multiple Agents 269
5.3 Ordering a Drink with Restricted Drink Choices 270
6 Plan Execution, Monitoring, and Recovery 271
7 Discussion and Conclusions 273
References 274

Produkt-Details

• Erscheint lt. Verlag: 25.3.2020
• Zusatzinfo: VIII, 277 p. 97 illus., 53 illus. in color.
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik ► Datenbanken
• Themenwelt: Wirtschaft ► Betriebswirtschaft / Management ► Wirtschaftsinformatik
• Schlagworte: AI Planning & Scheduling • Artificial Intelligence • Computer Science • knowledge based systems • Knowledge Capture • Knowledge Encoding • Knowledge Engineering • Knowledge Engineering Tools • Model-Based Reasoning • Validation & Verification
• ISBN-10: 3-030-38561-2 / 3030385612
• ISBN-13: 978-3-030-38561-3 / 9783030385613