Abstract:

This paper addressed the issues of insufficient safety distance, unsmooth path, and low planning efficiency in AGV path planning in narrow road scenarios. A new path planning method based on Voronoi skeletons for narrow roads was proposed. Firstly, the key nodes of the Voronoi skeleton were extracted to generate a custom Voronoi graph layer, which was then combined with static, obstacle, and expansion layers to iteratively generate a four-layer network structure cost map, accurately distinguishing the influence range of obstacles. Secondly, the new cost map was used as a constraint in conjunction with an improved A^* algorithm for global path planning, guiding the AGV to drive along the center of the road to ensure its safety. Finally, the global path was optimized using B-spline smoothing to improve the AGV's efficiency and stability in navigating narrow roads and other complex scenarios. Experimental results show that, in narrow road scenarios, the safety of the AGV path planning is improved by 82%, the number of spatial corners is reduced by 55.85%, and the path planning time is shortened by 48.98%. The proposed algorithm effectively enhances the robustness, dynamic obstacle avoidance, and real-time performance of path planning in narrow road scenarios, enabling the AGV to move in the safest manner.

Key words: Voronoi skeleton, narrow roads, automated guided vehicle (AGV), path planning, four-layer network structure, B-spline optimization