CAMPs: Learning Context-Specific Abstractions for Efficient Planning in Factored MDPs
arXiv:2007.13202
Abstract
Meta-planning, or learning to guide planning from experience, is a promising approach to improving the computational cost of planning. A general meta-planning strategy is to learn to impose constraints on the states considered and actions taken by the agent. We observe that (1) imposing a constraint can induce context-specific independences that render some aspects of the domain irrelevant, and (2) an agent can take advantage of this fact by imposing constraints on its own behavior. These observations lead us to propose the context-specific abstract Markov decision process (CAMP), an abstraction of a factored MDP that affords efficient planning. We then describe how to learn constraints to impose so the CAMP optimizes a trade-off between rewards and computational cost. Our experiments consider five planners across four domains, including robotic navigation among movable obstacles (NAMO), robotic task and motion planning for sequential manipulation, and classical planning. We find planning with learned CAMPs to consistently outperform baselines, including Stilman's NAMO-specific algorithm. Video: https://youtu.be/wTXt6djcAd4 Code: https://git.io/JTnf6
arXiv commentCoRL 2020
References in corpus (8)
- Decision-Theoretic Planning: Structural Assumptions and Computational Leverage
- Context-Specific Independence in Bayesian Networks
- SPUDD: Stochastic Planning using Decision Diagrams
- Near Optimal Behavior via Approximate State Abstraction
- Context-Specific Approximation in Probabilistic Inference
- Correlated Action Effects in Decision Theoretic Regression
- Think Too Fast Nor Too Slow: The Computational Trade-off Between Planning And Reinforcement Learning
- Proximity-Based Non-uniform Abstractions for Approximate Planning