Hybrid algorithm-based collision avoidance path planning for automated guided vehicles in intelligent warehouse

ObjectiveThis study proposes a two-stage path planning method in view of the path planning for automated guided vehicles in intelligent warehouse with static obstacles and dynamic obstacles. Safe and efficient operating paths for automated guided vehicles will be generated by using a hybrid algorithm fused with particle swarm optimization and improved dynamic window.MethodFirst, the particle swarm optimization algorithm was used to generate the global path with the objectives to minimizing the path length and driving risk. The key nodes selection strategy was used to remove redundant points, and retain the intermediate target points. Second, the dynamic window approach was improved by enhancing the parameters of evaluation function and changing the reference position of heading(v, ω) to reduce defects, e.g., generating path redundancy and unreachable target point. Finally, the particle swarm optimization algorithm and the improved dynamic window approach were fused. The sub-target points from the previous step were used as the process target points of the improved dynamic window approach to generate the final path. The automated guided vehicles could not only drive along the globally shortest path, but also had the ability to avoid the unknown obstacles in intelligent warehouse. Different warehouse environments were simulated by using Matlab to run simulation tests with the proposed algorithm.ResultThe improved algorithm generates smaller turning angle and higher path smoothness compared with the dynamic window approach. The path length is shortened by 5.12%, and the running time is reduced by 41.33%.ConclusionThe proposed algorithm could effectively complete the path planning and achieve the effective obstacle avoidance of dynamic obstacles, verifiing the effectiveness and feasibility of the algorithm.