Solving classical AI planning problems using planning-independent CP modeling and search

Abstract

The combinatorial problems that constraint programming typically solves belong to the class of NP-hard problems. The AI planning commun ity focuses on even harder problems: for example, classical planning is PSP ACE-hard. A natural and well-known constraint programming approach to classical planning solves a succession of fixed plan-length problems, though to date it has had limited success. We revisit this approach in light of recent progress on general-purpose branching heuristics. We conduct an empirical comparison to show the importance of using effective combinatorial search heuristics with this approach and that the quality of the plans produced is sometimes comparable to that of state-of-the-art planners.