A Macro-Control and Micro-Autonomy Pathfinding Strategy for Multi-Automated Guided Vehicles in Complex Manufacturing Scenarios

To effectively plan the travel paths of automated guided vehicles (AGVs) in complex manufacturing scenarios and avoid dynamic obstacles, this paper proposes a pathfinding strategy that integrates macro-control and micro-autonomy. At the macro level, a central system employs a modified A* algorithm for preliminary pathfinding, guiding the AGVs toward their targets. At the micro level, a distributed system incorporates a navigation and obstacle avoidance strategy trained by Prioritized Experience Replay Double Dueling Deep Q-Network with ε-Dataset Aggregation (PER-D3QN-EDAgger). Each AGV integrates its current state with information from the central system and the neighboring AGVs to make autonomous pathfinding decisions. The experimental results indicate that this strategy exhibits a strong adaptability to diverse environments, low path costs, and rapid solution speeds. It effectively avoids the neighboring AGVs and other dynamic obstacles, and maintains a high task completion rate of over 95% when the number of AGVs is below 200 and the obstacle density is below 0.5. This approach combines the advantages of centralized pathfinding, which ensures high path quality, with distributed planning, which enhances adaptability to dynamic environments.