A Study on Drone Logistics Delivery Based on Multi-Center Routing

With the rapid growth in e-commerce demand, increasing pressure on same-day delivery, and rising last-mile logistics costs, UAV-based logistics systems have emerged as a promising solution for efficient transportation in complex environments. In mountainous regions, however, irregular terrain, limited infrastructure accessibility, and strict flight constraints significantly increase the difficulty of logistics planning. To address these challenges, this study proposes a two-layer collaborative optimization framework for multi-center UAV logistics delivery systems. At the lower level, a multi-center site selection model was developed to determine the optimal distribution center locations and assign task areas. A trajectory cost matrix was constructed by comprehensively considering multiple constraints. The model was solved using a hybrid strategy that combines chaotic initialization and local enhancement based on the elite saDE method to improve the Starfish Optimization Algorithm, called the Mixed-Strategy Improved Starfish Optimization Algorithm (MISFOA), thereby generating feasible three-dimensional flight trajectories between local nodes. At the upper level, an improved Adaptive Large Neighborhood Search (IALNS) algorithm is applied to perform UAV mission assignment and route scheduling within each distribution center, based on the trajectory cost matrix pre-calculated at the lower level. The proposed framework achieves effective information exchange and hierarchical coupling between center selection and scheduling at the distribution level, thereby enabling unified optimization of the multi-center location and coordinated dispatch system. Simulation results demonstrate that the proposed method significantly improves delivery efficiency and solution quality in complex mountainous environments while ensuring trajectory feasibility and operational safety. This model provides a scalable and practical optimization framework for low-altitude logistics network planning under complex constraints.