Unmanned Aerial Vehicle Path Planning for Urban Emergency Supplies Delivery Based on Improved A* Algorithm

The paper presents a 3D obstacle avoidance path planning model for Unmanned Aerial Vehicles (UAVs) in complex urban low-altitude environments to address the issues encountered by the traditional A* algorithm, such as excessive inflection points, lengthy paths, and compromised safety. The model is designed by considering the spatial characteristics of building positions and heights in urban settings. Enhancements include refining the heuristic function of the A* algorithm and optimizing redundant corners. The A* algorithm is further refined by integrating the gradient descent algorithm and the Savitzky-Golay (S-G) filter. Comparative simulations involving the traditional Rapidly-exploring Random Tree (RRT) algorithm, the conventional A* algorithm, and the improved A* algorithm are conducted. Results demonstrate that the improved A* algorithm not only boosts search efficiency, maximum turning angle, and path length but also notably enhances the smoothness of Unmanned Aerial Vehicle (UAV) trajectories, showcasing superior overall performance. The algorithm effectively circumvents static obstacles, offering significant practical value in UAV track planning. In typical urban scenarios, the improved algorithm reduces the number of trajectory grids by 48.7% and decreases the maximum turning angle from 4.3165° to 0.9515°, showcasing its practical utility for engineering applications.