A real-time bi-level path planning approach for the conflict-free routing problem of multi-AGV systems

Abstract

A real-time bi-level path planning approach is proposed to address the conflict-free routing problem of multi-AGV systems. At the static path planning level, an enhanced A^* algorithm is proposed, which considers risks of different types of conflicts as path searching costs, in order to minimize conflicts during path searching. At the dynamic traffic control level, Petri net based traffic controllers are designed for dynamic traffic management on conflicting nodes and arcs to prevent AGV collisions. Computational test results demonstrate that the improved A^* algorithm can significantly reduce preplanned path conflicts. The Petri net based traffic controllers are able to effectively guide AGVs to avoid collisions when conflicts arise. The computational times for both levels are within milliseconds, which is crucial for real-time applications.