Most required goals for extracting heuristics in planning

A novel heuristic search planner, MRGFF (most required goals for fast forward planning), is advanced in this paper based on the well-known fast-forward planning system, MRGFF uses a notion, most required goals, to compute a lower heuristic estimate of a state and to prune the search space efficiently. The system has been implemented in C on a machine with a Pentium III processor at 600MHz and 160M memory, running Linux. Empirical results show the excellent performance achieved by our algorithm in terms of efficiency, scalability and accuracy.