AGV Traffic Management in Automated Industrial Plants: An Enhanced Lifelong Multi-Agent Path Finding Approach

In the context of Logistics 4.0, effective management of fleets of mobile agents in automated industrial plants is imperative to ensure productivity and enhance business flexibility. This study introduces an AGV traffic management system based on the Lifelong Multi-Agent Path Finding (L-MAPF) algorithm for the coordination of a fleet of AGVs. Unlike previous studies applied to a standardized grid-like environment, we focus on real-world industrial scenarios with bidirectional narrow corridors, where AGVs operate, moving on a roadmap. Our solution is based on a three-layer environment representation, i.e., a gridmap layer useful to represent the workspace of the AGVs, a roadmap layer employed to plan AGV paths, and a topological layer that divides the plant into different sectors, like corridors. For complete conflict resolution, we utilize our Bounded Horizon Conflict-Based Search, incorporating two innovative strategies for calculating corridor extension on the roadmap and expanding the time horizon within the corridor. The results of the tests conducted in a real industrial setting are presented and discussed in detail, demonstrating the effectiveness of our proposed method. Specifically, this methodology guarantees real-time AGV coordination and mitigates deadlock situations.