GRouteNet: A GNN-Based Model to Optimize Pathfinding and Smart Charging Management for Autonomous Guided Vehicles

Autonomous Guided Vehicles (AGVs) play an important role in the automation of material handling and transportation tasks in modern industrial and logistics systems. However, suboptimal path planning and longer waiting times at charging stations significantly affect the operational efficiency of these vehicles. To address these challenges, we leverage the capabilities of Graph Neural Networks (GNN) to find the optimal paths for AGVs. In this paper, we propose GRouteNet, a GNN-based model that effectively finds the shortest path for AGVs. The proposed model utilizes the message-passing mechanism of GNN to determine the neighbor nodes and then aggregates this information to find the shortest path. We compare the results of GRouteNet with a couple of existing state-of-the-art pathfinding models and show that the path length computed by GRouteNet is up to 45% shorter compared to the existing models. Furthermore, we propose a Shortest Charging Time First (SCTF) scheduling algorithm to reduce the long waiting times in the queues at the charging stations. The proposed SCTF algorithm prioritizes the charging of AGVs based on their charging time and charges the AGVs with the shortest charging time first. We compare the results of SCTF with the first-come-first-serve scheduling algorithm and show that SCTF reduces the waiting times at the charging stations by up to 42%.