Abstract: Efficient multi-agent path finding (MAPF) is essential for large-scale warehousing and logistics systems. Despite the potential of reinforcement learning (RL) methods, current approaches ...