关键词: 分区优化, 物流无人机, 无人机全自动机场, 选址, Voronoi图

Abstract: With the flourishing development of e-commerce and new retail, the demand for logistics and delivery has been continuously increasing. With the improvement of technology and policies, UAV urban delivery will become a trend in future development. Currently, it is being promoted and applied in cities such as Shenzhen and Hangzhou for express delivery and medical blood sample delivery, etc. With the special safety requirements in the background of the COVID-19 pandemic, the concept of “contactless delivery” has emerged. The fully automated vertiport is a completely contactless service, and UAV delivery expands the delivery range of instant delivery, improves delivery speed, and innovates urban delivery modes.
This paper studies the location selection problem of UAV urban delivery. Though the method of improved grid partitioning and the Voronoi polygon, optimizing the partition vertiport service area and the distribution of demand points is more balanced, and the goal of global optimization is achieved.
Firstly, this study identifies the UAV delivery operational scenarios and constructs a vertiport location model. The study focuses on automatic vertiport (Robort Hub) of Hangzhou Xunyi Network Technology Co., Ltd., which provides ground infrastructure supporting UAV vertical takeoff, landing, parking, charging, and maintenance. The location model for the vertiport is constructed by considering the factors affecting logistics UAV urban operations, the need for vertiport coverage the whole area, and the distance from the merchants to the vertiports. And the objective is to minimize the total operating time of the UAV vertiport delivery mode.
Next, the study of UAV urban delivery partition and vertiport location is conducted in Heping District of Tianjin City as an example. First, by crawling the block data of Heping District’s catering merchants and using the clip tool in ArcMap to clip the merchant data points, 2190 pieces of catering merchant data in Heping District are obtained. Second, based on the actual data of the merchants, an improved grid partition method is used to partition the demand point area. The Counting Cliques algorithm is used to find out the location of vertiport. Since the improved grid method may result in demand points being too concentrated in the partitioned areas and demand points in adjacent grid areas being closer to other area vertiports, the study proposes the use of the Voronoi polygon method to redistribute the demand points in each grid to ensure that the distance from the demand points within the area to the vertiports is minimized, and the distribution of demand points is more balanced, achieving the global optimal solution. Furthermore, the possibility of peak congestion during delivery is considered, and the “anti-mobility” index is analyzed to determine whether to increase the number of vertiports in a given area.
Finally, though the application of the model and the partitioning methods, the results of optimal partition and vertiport location in Heping District of Tianjin is obtained.
Using an improved grid partitioning method, the demands of merchants in the Heping District are partitioned into 74 effective grid partitions. Based on the location model, the vertiport location is determined. By using the Voronoi polygon method, demand points are reassigned to achieve a more even distribution. After analyzing the “anti-mobility” index, it is found that adding new UAV vertiports in areas with higher metric values would be beneficial and could accelerate logistics speed.
The partition optimization method allows for a more reasonable service area for the siting of UAV vertiport, this method is also applicable to other urban areas suitable for UAV delivery, providing a basis for ground facilities and equipment support for Urban Air Mobility (UAM) systems.
In this study, the service area for siting UAV vertiports is optimized. Although the problem of local optimization has been solved, when determining the final location of the UAV vertiport, it is necessary to further consider factors such as obstacles, building heights, distance from buildings, and energy consumption of UAVs in the actual environment. In addition, the impact of actual demand of urban UAV delivery on the location and number of vertiports should be explored.

Key words: partition optimization, logistics Unmanned Aerial Vehicle (UAV), vertiport, location, Voronoi polygon