Type-based exploration with multiple search queues for satisficing planning

摘要: Utilizing multiple queues in Greedy Best-First Search (GBFS) has been proven to be a very effective approach satisficing planning. Successful techniques include extra based on Helpful Actions (or Preferred Operators), as well using Multiple Heuristics. One weakness of all standard GBFS algorithms is their lack exploration. All used these methods work priority sorted by heuristic values. Therefore, misleading heuristics, especially early the search process, can cause become ineffective. Type systems, introduced for Lelis et al, are development ideas exploration related classic stratified sampling approach. The current introduces algorithm that utilizes type systems new way-for within multiqueue framework planning. A careful case study shows benefits such overcoming deficiencies heuristic. proposed baseline Type-GBFS solves almost 200 more problems than over International Planning Competition problems. Type-LAMA, planner which integrates into LAMA-2011, 36.8 LAMA-2011.