Representation-Optimal Multi-Robot Motion Planning Using Conflict-Based Search

Multi-Agent Motion Planning (MAMP) is the problem of computing feasible paths for a set agents each with individual start and goal states within continuous state space. Existing approaches can be split into coupled methods which provide optimal solutions but struggle scalability or decoupled scalable offer no optimality guarantees. Recent work has explored hybrid that leverage advantages both in an easier discrete subproblem, Pathfinding (MAPF). In this work, we adapt recent developments MAPF to domain MAMP. We demonstrate our method manage groups up 32 agents, ability handle 8 high-DOF manipulators, plan heterogeneous teams. all scenarios, approach plans significantly faster while providing higher quality solutions.