In addressing the time-consuming and computationally intensive nature of finding optimal solutions for unmanned aerial vehicle (UAV) path planning in urban logistics applications, an improved A* algorithm was proposed. Firstly, a three-dimensional environmental model was established using grid-based methods, and a multi-constrained logistics UAV path planning model was constructed based on UAV performance, transforming the path planning problem into an optimal path search problem in three-dimensional space. Secondly, the improved A* algorithm incorporates the concept of dynamic step size, Bresenham algorithm principles, and an enhanced heuristic function to enhance the quality and efficiency of the algorithm's path planning. Finally, by introducing diagonal distance, Euclidean distance, and Manhattan distance as the heuristic function for the improved A* algorithm, and combining it with the established three-dimensional environmental model, a parameter search method was used to calculate the optimal weight values, thus proposing an efficient optimization algorithm for three-dimensional path planning of urban logistics UAVs. Case studies indicate that the average time for path planning using the improved A* algorithm is reduced by at least 24.84% compared to traditional A* and JPS algorithms, the average path length is shortened by at least 2.89%, and the average number of search nodes is reduced by at least 82.81%.