Practical multi-agent path finding (MAPF) for lifelong operation in warehouse and industrial environments requires not only completeness and computational efficiency, but also robustness to disturbances and failures. This paper addresses robust and complete replanning in prioritized MAPF. We extend a previously proposed lifelong MAPF framework with a formalized strategy to handle unexpected disturbances such as delays, road blockages, and agent failures. The proposed approach combines safe locations, priority ordering, and relocation mechanisms to guarantee recovery whenever a feasible solution exists, while identifying unrecoverable agents otherwise
|
