Multi-Agent Path Finding - An Overview

Multi-Agent Pathfinding (MAPF) is the problem of finding paths for multiple agents such that every agent reaches its goal and the agents do not collide. In recent years, there has been a growing interest in MAPF in the Artificial Intelligence (AI) research community. This interest is partially because real-world MAPF applications, such as warehouse management, multi-robot teams, and aircraft management, are becoming more prevalent. In this overview, we discuss several possible definitions of the MAPF problem. Then, we survey MAPF algorithms, starting with fast but incomplete algorithms, then fast, complete but not optimal algorithms, and finally optimal algorithms. Then, we describe approximately optimal algorithms and conclude with non-classical MAPF and pointers for future reading and future work.