Abstraction heuristics for symbolic bidirectional search

摘要: Symbolic bidirectional uniform-cost search is a prominent technique for cost-optimal planning. Thus, the question whether it can be further improved by making use of heuristic functions raises naturally. However, heuristics in does not always improve its performance. We propose novel way to abstraction symbolic which only resorts when becomes unfeasible. adapt definition partial and perimeter abstractions search, where A* used traverse abstract state spaces and/or generate perimeter. The results show that some domains. In fact, resulting planner, SymBA*, was winner optimal-track last IPC.