Efficient symbolic search for cost-optimal planning

摘要: In cost-optimal planning we aim to find a sequence of operators that achieve set goals with minimum cost. Symbolic search Binary Decision Diagrams (BDDs) performs efficient state space exploration in terms time and memory. This is crucial optimal settings, which large parts the must be explored order prove optimality. However, development accurate heuristics for explicit-state recent years have left symbolic techniques secondary place.In this article propose two orthogonal improvements planning. On one hand, analyze compare different methods image computation efficiently perform successor generation on search. Image main bottleneck algorithms so an paramount other study how use state-invariant constraints prune states essential regression but it yet exploited planners.Experiments bidirectional uniform-cost A * PDBs show remarkable performance most IPC benchmark domains. Overall, help our improvements, outperforms state-of-the-art such as LM-cut across many