依据基于动作的编码方式,提出一种基于约简动作变元的自动命题规划编码方式:基于命题的编码方式。首先分析构造新编码方式的理论依据,提出基于命题的编码方式的编码组成,证明其有效性,并描述某些公理的具体实现细节,最后分析其与已有几种编码方式的不同之处。在SATPLAN2006规划系统中实现了基于命题的编码方式,利用国际规划竞赛选用的Benchmark问题予以测试,分析其与基于动作的编码方式等2种极端的编码方式的求解特性。实验结果表明:对于顺序规划问题域,基于命题的编码方式更有效;对于并发规划问题域,基于动作的编码方式更有效。
Based on action-based encoding, it proposes another automated propositional planning encoding by reducing action variables, called proposition-based encoding. It firstly analyzes the theoretical foundations to propose a new encoding, proposes the composition of proposition-based encoding, justifies the soundness and completeness of the corresponding encoding, and also describes the feasible actual encoding methods for some axioms. Finally, it analyzes the distinctions between it and existed encodings. It implements proposition-based encoding in SATPLAN2006 planner, and tests it and action-based encoding using benchmarks adopted by Intemational Planning Competition to analyze their specialties. The experimental results validate that for sequential planning problems Blocks World, proposition-based encoding is superior to action-based encoding, and for parallel planning problems Logistics, the latter is superior to the former.