Practical Planning

Extending the Classical AI Planning Paradigm


  • David Wilkins

Planning, or reasoning about actions, is a fundamental element of intelligent behavior--and one that artificial intelligence has found very difficult to implement. The most well-understood approach to building planning systems has been under refinement since the late 1960s and has now reached a level of maturity where there are good prospects for building working planners.

Practical Planning is an in-depth examination of this classical planning paradigm through an intensive case study of SIPE, a significantly implemented planning system. The author, the developer of SIPE, defines the planning problem in general, explains why reasoning about actions is so complex, and describes all parts of the SIPE system and the algorithms needed to achieve efficiency. Details are discussed in the context of problems and important issues in building a practical planner; discussions of how other systems address these issues are also included.

Assuming only a basic background in AI, Practical Planning will be of great interest to professionals interested in incorporating planning capabilities into AI systems.

View full description


Book information

  • Published: September 1988
  • ISBN: 978-0-934613-94-1

Table of Contents

Practical Planning: Extending the Classical AI Planning Paradigm
David E. Wilkins
List of Figures
List of Tables
1 Reasoning about Actions and Planning1.1 Philosophical and Practical Importance
1.2 The Classical AI Planning Problem
1.3 Reactive Planning
1.4 The Essence of Planning
1.5 Capabilities of a Planning System
1.6 How Hard is Planning?
1.7 Classical AI Planning Systems
1.8 SIPE
2 Basic Assumptions and Limitations2.1 Important Features
2.2 Limitations
3 SIPE and Its Representations3.1 Representation of Domain Objects and Relationship
3.2 Operator Description Language
3.3 Plan Rationale
3.4 Plans
4 Hierarchical Planning as Differing Abstraction Levels4.1 The Many Guises of Hierarchical Planning
4.2 A Problem with Current Planners
4.3 Solutions
5 Constraints5.1 SIPE's Constraint Language
5.2 Use of Constraints
5.3 Unification
6. The Truth Criterion6.1 The Formula Truth Criterion
6.2 The PTC for Ground, Linear Plans
6.3 Introducing Variables
6.4 Introducing Existential Quantifiers
6.5 Introducing Universal Quantifiers
6.6 Introducing Nonlinearity
6.7 Summary
7 Deductive Causal Theories7.1 A Motivating Example
7.2 Domain Rules
7.3 problems with Domain Rules
7.4 Heuritic Adequacy and Expressive Power
8 Plan Critics8.1 Solving the Constraint Network
8.2 Parallel Interactions
8.3 Goal Phantomization
8.4 Solving Harmful Interactions
8.5 Adding Ordering Constraints
8.6 Examples
9 Resources: Reusable, Consumable, Temporal9.1 Reusable Resources
9.2 Representation of Numerical Quantities
9.3 Consumable Resources
9.4 Temporal Reasoning
9.5 Manipulating Numerical Quantities
9.6 Summary
10 Search10.1 Automatic Search
10.2 Intermingling Planning and Execution
10.3 Interactive Control
10.4 Domain-Dependent Search Control
10.5 Other Search Strategies
11 Replanning During Execution11.1 Overview of SIPE's Execution-Monitoring System
11.2 Unknowns
11.3 Interpreting the Input
11.4 The Problem Recognizer
11.5 Replanning Actions
11.6 The General Replanner
11.7 Examples
11.8 Searching the Space of Modified Plans
11.9 Summary
12 Planning and Reactivity12.1 Level of the Interface
12.2 Who is in Control?
13 Achieving Heuristic Adequacy13.1 Summary of Heuristics
13.2 Subsumption of Pred Constraints
13.3 Encoding Domains in SIPE
14 Comparison with Other Systems14.1 Nonclassical Planning Systems
14.2 Previous Classical Planners
14.3 Constraints
14.4 Critics
14.5 Replanning
14.6 Heuristic AdequacyBibliography