Memory-based heuristics for explicit state spaces

摘要: In many scenarios, quickly solving a relatively small search problem with an arbitrary start and goal state is important (e.g., GPS navigation). order to speed this process, we introduce new class of memory-based heuristics, called true distance that store distances between some pairs states in the original space can be used for heuristic any pair states. We provide number techniques using improving heuristics such most benefits all-pairs shortest-path computation gained less than 1% memory. Experimental results on domains show 6- 14 fold improvement compared traditional heuristics.