Improvement of Task Allocation in TP Algorithm for MAPD Problem by Relaxation of Movement Limitation and Estimation of Pickup Time

Multi-Agent Path Finding (MAPF) problem is an extended class of pathfinding problems where multiple paths are assigned to agents without collision. MAPF can been applied various practical domains including path-planning robot carriers in warehouses, automated taxiing airplanes, and video games. Multi Agent Pickup Delivery (MAPD) that a continuous has studied for more cases such as warehouses which tasks appear over time. A major goal MAPD reduce the execution time or total path length tasks. Token Passing (TP) proposed solution method problems. TP task allocation also employs algorithm Cooperative A* (CA*) Multi-Label (MLA*) algorithm. With TP, pickup-and-delivery sequentially allocated by employing shared memory called token reserve their In this study, we focus on two issues TP. Firstly, targets well-formed satisfy sufficient condition solvable instances. The MAPDs limit passable nodes graph representation warehouse decreases utilization space pathfinding. However, if restriction relaxed, collision between agents’ might occur. Secondly, greedily performed depending default order unallocated access token. there be another agent whose current almost completed its destination nearer pickup location new task. Without considering agent, it generate redundant allocating inappropriate agent. To address first problem, relax movement limitation introduce reservation stays at destinations maintains consistency reserved agents. For second propose estimates determine appropriate Experimental results show our reduces reducing comparison existing method.