UAV Path Planning Based on the DCE-A* Algorithm

As unmanned aerial vehicles (UAVs) are increasingly deployed in complex three-dimensional environments, the traditional A* algorithm with 26-neighborhood expansion suffers from low search efficiency, redundant paths, and heavy computational burden, making it difficult to simultaneously satisfy the real-time and feasibility requirements of path planning. In this paper, a direction-cost-enhanced A* (DCE-A*) algorithm based on direction indexing and expansion pruning is proposed. A directional scoring mechanism is introduced to preferentially expand intermediate nodes oriented toward the target, and a Top-K node selection strategy is employed in the expansion stage to retain only the K_\mathrmtop nodes with the highest directional scores, thereby reducing redundant expansions. In the post-processing stage, a high-order Bézier curve is used for global path fitting to improve path continuity and flight controllability. Simulation experiments on multi-scale 3D grid maps show that, while ensuring path feasibility, the proposed algorithm reduces the total number of visited nodes by 46%～66%, decreases planning time by 45%～67%, increases the success rate of path search by about 38%, and shortens the path length by 6.45% in complex scenarios compared with the conventional A* algorithm.