Improve Multi-agent Path Finding by Bridging the Gap between Abstract Algorithms and Specific Application Scenarios

The movement of mobile robots in a known environment can be viewed as a Multi-Agent Path Finding (MAPF) problem. MAPF is influenced by various factors such as the application scenarios, the scale and attributes of the agents. However, MAPF assumes agents to be abstract homogeneous nodes, without considering differences in agent attributes. Therefore, this paper proposes a general method to improve the abstract MAPF algorithms into the specific Multi-Attribute Heterogeneous MAPF (MAH-MAPF) algorithms. The method consists of three steps: 1) Define the MAH-MAPF problem based on the attributes of the robots in the application scenario. 2) Based on the agent attributes and definition, transform the MAPF algorithm into MAH-MAPF algorithm by modifying the homogeneous abstract agent assumption. 3) Reduce path conflicts caused by agent attributes to improve the performance of the MAH-MAPF algorithm. The experimental results indicate that the method proposed in this paper is effective and general.