Parallel Strategies for Best-First Generalized Planning
Alejandro Fernández-Alburquerque, Javier Segovia-Aguas·July 31, 2024
Summary
The paper focuses on parallelizing Best-First Generalized Planning (BFGP), a heuristic-based approach for solving multiple planning instances from the same domain. BFGP uses novel heuristics and a new solution space, independent of the number of input instances, to compute general algorithmic solutions. The authors propose two shared-memory parallel strategies that scale linearly with the number of cores, aiming to bridge the performance gap between state-of-the-art planning solvers and generalized planning. The first strategy sequentially expands nodes until there are at least N nodes per thread, then starts a parallel search where each thread is independent. The second strategy distributes promising nodes during the parallel search phase, balancing the tradeoff between searching promising states and minimizing communication overhead. The results show that BFGP is well-suited for parallelization, with potential for handling more complex problems from IPC planning domains. The paper also discusses various aspects of generalized planning, including methods for efficient navigation of the planning space, representation and synthesis of C++ programs, and a new planning search-space that leverages pointers over objects. The 2023 International Planning Competition highlights advancements and challenges in the field of generalized planning.
In summary, the paper presents parallelization strategies for the Best-First Generalized Planning algorithm, demonstrating its suitability for handling complex planning problems and closing the performance gap between state-of-the-art planning solvers and generalized planning. The work contributes to the field of AI research focused on automated synthesis of algorithmic-like solutions for multiple planning instances, with potential for advancements in efficient navigation, representation, and synthesis of planning solutions.
Introduction
Background
Overview of Best-First Generalized Planning (BFGP)
Importance of heuristic-based approaches in planning
Novel heuristics and solution space in BFGP
Objective
Aim of the research: bridging the performance gap between planning solvers and generalized planning
Focus on parallel strategies for BFGP
Method
Parallel Strategies for BFGP
Shared-Memory Parallel Strategies
Strategy 1: Sequential node expansion followed by parallel search
Strategy 2: Distributed promising nodes during parallel search
Scalability and Performance
Linear scalability with the number of cores
Handling of complex planning problems
Implementation Details
C++ programs for representation and synthesis
Planning search-space leveraging pointers over objects
Results
Performance Evaluation
Comparison with state-of-the-art planning solvers
Potential for more complex problem handling
Case Studies
Examples from the 2023 International Planning Competition
Discussion
Generalized Planning Aspects
Efficient navigation of the planning space
Representation and synthesis of planning solutions
Challenges and Future Directions
Advancements and challenges in generalized planning
Potential for further research and development
Conclusion
Summary of Contributions
Parallelization strategies for BFGP
Suitability for complex planning problems
Closing the performance gap between planning solvers and generalized planning
Implications for AI Research
Automated synthesis of algorithmic-like solutions
Potential for advancements in planning algorithms and techniques
Basic info
papers
artificial intelligence
Advanced features