Automated Roadmap Graph Creation and MAPF Benchmarking for Large AGV Fleets

In this paper we present a concept and first evaluation of an automated topology creation and two different Multi-Agent Path Finding (MAPF) approaches for large robot fleets with the focus on logistic applications. Based on a map or a CAD file and the robot size an algorithm for creating a so called Roadmap Graph will be introduced. The algorithm is able to utilize as much of the available space as possible and can also guarantee that neither edges nor vertices overlap each other which may lead to collisions of the robots. In order to analyse the quality of the Roadmap Graph, an analysis of different graph variants using measures from graph theory is presented. Additionally, an evaluation of routing performance using one of the MAPF algorithms has been conducted on these graphs. Finally, a brief evaluation and comparison of the two MAPF algorithms on exemplaric gridmap graphs show the generic advantages and disadvantage of both algorithms.