Multi-Agent Path Finding with Time Windows: Preliminary Results
Abstract
We formalize the problem of multi-agent path finding with time windows (MAPF-TW). The optimization objective is to maximize the average customer satisfaction for all agents when they reach their respective goal vertices without path conflicts. We first prove that solving MAPF-TW optimally is NP-hard. We then reduce the MAPF-TW problem into a multi-commodity flow problem and propose an integer linear programming (ILP) model. Next, we propose the conflict-based search with time windows (CBS-TW) for the MAPF-TW problem, which is also optimal. Finally, we conduct simulation experiments on two different maps with random obstacles.