Encoding Factored Tasks in SAT: Strategies and Impact of Transformations
This research explores the encoding of factored tasks, a traditional planning model that enhances SAS+, into SAT (Boolean satisfiability). Factored tasks facilitate a concise representation through disjunctive preconditions, conditional effects, and angelic nondeterminism; however, current planners rely solely on heuristic search methods. The authors suggest several strategies for converting the factored transition relation into propositional logic. They examine ways to leverage parallelism across different levels and assess how typical task transformations impact the performance of SAT-based planners. The objective of this study is to broaden the application of SAT-based planning to encompass more expressive formalisms.
Key facts
- Factored tasks extend SAS+ with disjunctive preconditions, conditional effects, and angelic nondeterminism.
- Existing planning approaches for factored tasks are limited to heuristic search.
- This work investigates encoding factored tasks in SAT.
- Multiple encoding strategies are proposed for translating the factored transition relation into propositional logic.
- The paper analyzes exploiting parallelism at various levels.
- The impact of common task transformations on SAT-based planner performance is studied.
- The goal is to enable SAT-based planning for factored tasks.
- The paper is an extended version, submitted to arXiv.
Entities
Institutions
- arXiv