Cost-Optimal Planning with Landmarks

Recently, Richter et al. [2008] proposed a novel some point in every solution plan. Previous work way of using a set of landmarks as a pseudo-heuristic within has very successfully exploited planning landmarks a satisficing heuristic search. This technique allowed both in satisficing (non-optimal) planning. We propose a reducing the length of the generated plans, as well as improving methodology for deriving admissible heuristic estimates success rate both with respect to the iterative approach for cost-optimal planning from a set of planning of Hoffmann et al., and with respect to other stateof-the-art landmarks. The resulting heuristics fall into a satisficing planners. In particular, the LAMA planner novel class of multi-path dependent heuristics, and by Richter and Westphal utilizing such a landmarks-based we present a simple best-first search procedure exploiting heuristic search was the clear winner of the Sequential Satisficing such heuristics.