Hybrid-Map Multi-AGF System for Truck Loading

This paper presents a multi-AGF (Automatic Guided Forklift) system for container truck loading in logistic scenarios. The system is designed to coordinate a group of AGFs to move cargo from the warehouse or platform to the container truck, and load them into the truck in a time-efficient manner. To adapt to inconsistent docking positions and orientations of the container truck, the system employs a hybrid map, which combines a static map and a dynamic local map, with the former containing a pre-built point cloud and a pre-designed path layout for the warehouse and the platform, and the latter an auto-built point cloud and an auto-generated path layout for the container truck. Each time a container truck arrives, the system builds a new dynamic local map, combining it with the static map, and then coordinating the AGFs to finish the loading tasks for the truck within the combined hybrid map. Coordination of the AGFs is achieved by a time window MAPF algorithm based on the path-segment conflict model to ensure deadlock avoidance. Real-world deployment results validate the feasibility and performance in accuracy and efficiency of the system.