Path Planning for Mobile Robot Based on the Fusion Algorithm of Improved A* and APF

Abstract

It's vital to study path planning in complex mining environments while using the AGV to deploy dangerous tasks. However, global path planning can't avoid real-time obstacles, and local path planning is prone to fall into local optimality, so a deep study is necessary for the mining AGV. The A* algorithm, a classical pathing planning, has problems such as redundant path search nodes, dangerous diagonal movements, and unsmooth paths, the Artificial Potential Field (APF) algorithm faces issues of local optima and unreachable target. Therefore, a hybrid heuristic function incorporating Euclidean distance and deflection angle cost is introduced, and a 5-neighborhood search method is prioritized to improve the search efficiency of the A* algorithm. Additionally, the algorithm is optimized to remove dangerous diagonal movement nodes. The repulsive potential field function of the APF algorithm is modified by adding the n-th power of the Euclidean distance from the robot to the target point to ensure the target point is reachable, and the robot's movement direction algorithm is improved to solve some local optima problems. However, single path planning methods often struggle with limitations, so this paper proposes an enhanced fusion path planning method based on improvements to the A* algorithm and the APF algorithm. The optimal nodes generated by the improved A* algorithm are used as local target points for the improved APF algorithm, allowing it to escape local optima and simultaneously addressing the unsmooth path issue of the improved A* algorithm. Finally, simulation experiments comparing the proposed method in different scenarios were conducted. The experimental results show that this method not only combines the advantages of the improved A* algorithm and the improved APF algorithm but also effectively achieves path planning in complex mining environments.