Multi-agent path finding with mutex propagation

Mutex propagation is a form of efficient constraint popularly used in AI planning to tightly approximate the reachable states from given state. We utilize this idea context Multi-Agent Path Finding (MAPF). When adapted MAPF, mutex provides stronger constraints for conflict resolution CBS, popular optimal search-based MAPF algorithm, as well MDD-SAT, an satisfiability-based algorithm. CBS with ability break symmetries and MDD-SAT make inferences than unit propagation. While existing work identifies limited requires manual design symmetry-breaking constraints, more general allows automated constraints. Our experimental results show that capable outperforming CBSH-RCT, state-of-the-art variant respect success rate. also often performs better