Understanding planning tasks : domain complexity and heuristic decomposition /

Understanding planning tasks : domain complexity and heuristic decomposition /

Saved in:
Bibliographic Details
Author / Creator:Helmert, Malte.
Edition:1st ed.
Imprint:Berlin ; New York : Springer, ©2008.
Description:1 online resource (xiv, 270 pages) : illustrations.
Language:English
Series:LNCS sublibrary. SL 7, Artificial intelligence
Lecture notes in computer science, 0302-9743 ; 4929. Lecture notes in artificial intelligence
LNCS sublibrary. SL 7, Artificial intelligence.
Lecture notes in computer science ; 4929.
Lecture notes in computer science. Lecture notes in artificial intelligence.
Subject:
Artificial intelligence.
Planning -- Mathematical models.
Problem solving -- Mathematical models.
Computer algorithms.
Heuristic programming.
Artificial intelligence.
Computer algorithms.
Heuristic programming.
Planning -- Mathematical models.
Problem solving -- Mathematical models.
Format: E-Resource Book
URL for this record:http://pi.lib.uchicago.edu/1001/cat/bib/11066796
Hidden Bibliographic Details
Other uniform titles:Helmert, Malte. Solving planning tasks in theory and practice.
ISBN:9783540777229
3540777229
9783540777236
3540777237
Notes:Includes bibliographical references and index.
Summary:Action planning has always played a central role in Artificial Intelligence. Given a description of the current situation, a description of possible actions and a description of the goals to be achieved, the task is to identify a sequence of actions, i.e., a plan that transforms the current situation into one that satisfies the goal description. This monograph is a revised version of Malte Helmert's doctoral thesis, Solving Planning Tasks in Theory and Practice, written under the supervision of Professor Bernhard Nebel as thesis advisor at Albert-Ludwigs-Universität Freiburg, Germany, in 2006. The book contains an exhaustive analysis of the computational complexity of the benchmark problems that have been used in the past decade, namely the standard benchmark domains of the International Planning Competitions (IPC). At the same time, it contributes to the practice of solving planning tasks by presenting a powerful new approach to heuristic planning. The author also provides an in-depth analysis of so-called routing and transportation problems. All in all, this book will contribute significantly to advancing the state of the art in automatic planning.
Other form:Print version: Helmert, Malte. Understanding planning tasks. 1st ed. Berlin ; New York : Springer, ©2008 9783540777229 3540777229
Standard no.:10.1007/978-3-540-77723-6
Table of Contents:
  • Part I. Planning Benchmarks
  • 1. The Role of Benchmarks
  • 1.1. Evaluating Planner Performance
  • 1.1.1. Worst-Case Evaluation
  • 1.1.2. Average-Case Evaluation
  • 1.2. Planning Benchmarks Are Important
  • 1.3. Theoretical Analyses of Planning Benchmarks
  • 1.3.1. Why Theoretical Analyses Are Useful
  • 1.3.2. Published Results on Benchmark Complexity
  • 1.4. Standard Benchmarks
  • 1.5. Summary and Overview
  • 2. Defining Planning Domains
  • 2.1. Optimization Problems
  • 2.1.1. Minimization Problems
  • 2.1.2. Approximation Algorithms
  • 2.1.3. Approximation Classes
  • 2.1.4. Reductions
  • 2.2. Formalizing Planning Domains
  • 2.3. General Results and Reductions
  • 2.3.1. Upper Bounds
  • 2.3.2. Shortest Plan Length
  • 2.3.3. Approximation Classes of Limited Interest
  • 2.3.4. Relating Planning and (Bounded) Plan Existence
  • 2.3.5. Generalization and Specialization
  • 3. The Benchmark Suite
  • 3.1. Defining the Competition Domains
  • 3.2. The Benchmark Suite
  • 3.2.1. IPC1 Domains
  • 3.2.2. IPC2 Domains
  • 3.2.3. IPC3 Domains
  • 3.2.4. IPC4 Domains
  • 3.3. Domains and Domain Families
  • 4. Transportation and Route Planning
  • 4.1. Transport and Route
  • 4.1.1. The Transport Domain
  • 4.1.2. The Route Domain
  • 4.1.3. Special Cases and Hierarchy
  • 4.2. General Results
  • 4.3. Plan Existence
  • 4.4. Hardness of Optimization
  • 4.5. Constant Factor Approximation
  • 4.6. Hardness of Constant Factor Approximation
  • 4.7. Summary
  • 4.8. Beyond Transport and Route
  • 5. IPC Domains: Transportation and Route Planning
  • 5.1. Gripper
  • 5.2. Mystery and Mystery Prime
  • 5.3. Logistics
  • 5.4. Zenotravel
  • 5.5. Depots
  • 5.6. Miconic-10
  • 5.7. Rovers
  • 5.8. Grid
  • 5.9. Driverlog
  • 5.10. Airport
  • 5.11. Summary
  • 6. IPC Domains: Others
  • 6.1. Assembly
  • 6.2. Blocksworld
  • 6.3. Freecell
  • 6.4. Movie
  • 6.5. Pipesworld
  • 6.6. Promela
  • 6.7. PSR
  • 6.8. Satellite
  • 6.9. Schedule
  • 6.10. Summary
  • 7. Conclusions
  • 7.1. Ten Conclusions
  • 7.2. Going Further
  • Part II. Fast Downward
  • 8. Solving Planning Tasks Hierarchically
  • 8.1. Introduction
  • 8.2. Related Work
  • 8.2.1. Causal Graphs and Abstraction
  • 8.2.2. Causal Graphs and Unary STRIPS Operators
  • 8.2.3. Multi-Valued Planning Tasks
  • 8.3. Architecture and Overview
  • 9. Translation
  • 9.1. PDDL and Multi-valued Planning Tasks
  • 9.2. Translation Overview
  • 9.3. Normalization
  • 9.3.1. Compiling Away Types
  • 9.3.2. Simplifying Conditions
  • 9.3.3. Simplifying Effects
  • 9.3.4. Normalization Result
  • 9.4. Invariant Synthesis
  • 9.4.1. Initial Candidates
  • 9.4.2. Proving Invariance
  • 9.4.3. Refining Failed Candidates
  • 9.4.4. Examples
  • 9.4.5. Related Work
  • 9.5. Grounding
  • 9.5.1. Overview of Horn Exploration
  • 9.5.2. Generating the Logic Program
  • 9.5.3. Translating the Logic Program to Normal Form
  • 9.5.4. Computing the Canonical Model
  • 9.5.5. Axiom and Operator Instantiation
  • 9.6. Multi-valued Planning Task Generation
  • 9.6.1. Variable Selection
  • 9.6.2. Converting the Initial State
  • 9.6.3. Converting Operator Effects
  • 9.6.4. Converting Conditions
  • 9.6.5. Computing Axiom Layers
  • 9.6.6. Generating the Output
  • 9.7. Performance Notes
  • 9.7.1. Relative Performance Compared to MIPS Translator
  • 9.7.2. Absolute Performance
  • 10. Knowledge Compilation
  • 10.1. Overview
  • 10.2. Domain Transition Graphs
  • 10.3. Causal Graphs
  • 10.3.1. Acyclic Causal Graphs
  • 10.3.2. Generating and Pruning Causal Graphs
  • 10.3.3. Causal Graph Examples
  • 10.4. Successor Generators and Axiom Evaluators
  • 10.4.1. Successor Generators
  • 10.4.2. Axiom Evaluators
  • 11. Search
  • 11.1. Overview
  • 11.2. The Causal Graph Heuristic
  • 11.2.1. Conceptual View of the Causal Graph Heurstic
  • 11.2.2. Computation of the Causal Graph Heuristic
  • 11.2.3. States with Infinite Heuristic Value
  • 11.2.4. Helpful Transitions
  • 11.3. The FF Heuristic
  • 11.4. Greedy Best-First Search in Fast Downward
  • 11.4.1. Preferred Operators
  • 11.4.2. Deferred Heuristic Evaluation
  • 11.5. Multi-heuristic Best-First Search
  • 11.6. Focused Iterative-Broadening Search
  • 12. Experiments
  • 12.1. Experiment Design
  • 12.1.1. Benchmark Set
  • 12.1.2. Experiment Setup
  • 12.1.3. Translation and Knowledge Compilation vs. Search
  • 12.2. Strips Domains from IPC1-3
  • 12.3. ADL Domains from IPC1-3
  • 12.4. Domains from IPC4
  • 12.5. Conclusions from the Experiment
  • 13. Discussion
  • 13.1. Summary
  • 13.2. Major Contributors
  • 13.2.1. Multi-valued Representations
  • 13.2.2. Task Decomposition Heuristics
  • 13.3. Minor Contributions
  • 13.4. Going Further
  • References
  • Index

Similar Items