Abstract: Deadlock is always a challenging problem for multi-agent pathfinding, especially when the system is in high scales in terms of number of agents and map size. Some recent studies showed that ...