Guided Symbolic Universal Planning

Symbolic universal planning based on the reduced Ordered Binary Decision Diagram (OBDD) has been shown to be an efficient approach for planning in non-deterministic domains. To date, however, no guided algorithms exist for synthesizing universal plans. In this paper, we introduce a general approach for guiding universal planning based on an existing method for heuristic symbolic search in deterministic domains. We present three new sound and complete algorithms for best-first strong, strong cyclic, and weak universal planning. Our experimental results show that guiding the search dramatically can reduce both the computation time and the size of the generated plans.