MAXSAT Heuristics for Cost Optimal Planning

The cost of an optimal delete relaxed plan, known as h+, is a powerful admissible heuristic but in general intractable to compute. In this paper we examine the problem computing h+ by encoding it MAXSAT problem. We develop new that utilizes constraint generation support computation sequence increasing lower bounds on h+. show close connection between computations performed recent approach for solving and hitting set recently proposed Using observe our can be initialized with landmarks computed LM-cut. By judicious use along technique lazy evaluation obtain speedups finding plans over LM-cut number domains. Our enables exploitation continued progress solving, also makes possible consider or approximating heuristics are even more informed by, example, adding some information about deletes back into encoding.