Symmetry Breaking: Satisficing Planning and Landmark Heuristics ICAPS-2013 Doctoral Consortium

In heuristic search planning, state-space symmetries are mostly ignored by both the search algorithm and the heuristic guidance. Only recently, Pochter, Zohar, and Rosenschein (2011) introduced an effective framework for detecting and accounting for state symmetries within A∗ cost-optimal planning. We extend this framework in order to allow usage of strictly larger symmetry groups. This enables more efficient pruning of strictly larger parts of the search space. Our approach is based on exploiting information about the parts of the transition system being gradually revealed by the search algorithms, such as A∗, WA∗ or even GBFS. In our work we will expand the use of cost-optimal planning by Pochter, Zohar, and Rosenschein (2011) to forward statespace search planning, and present two implementations of this framework using different pruning algorithms. Our aim is to evaluate their effectiveness both theoretically and prac-