Real-time order dispatch for on-demand food-delivery platforms with a mixed fleet of drones and couriers

This paper investigates a dynamic, multimodal delivery system where a platform coordinates a mixed fleet of drones and human couriers to provide real-time delivery services. Launchpads serve as hubs, enabling order transferring between couriers and drones. Drones operate exclusively between launchpads, while couriers handle first- and last-mile deliveries with order bundling, either directly completing orders or delivering them to or from launchpads. The platform determines order delivery mode, assigns couriers to orders, and repositions idle drones to minimize total operational costs while meeting order delivery deadlines. We formulate this as a mixed-integer linear program (MILP), capturing seamless coordination among couriers and drones under a re-optimization policy. A novel decomposition method integrating graph-based methods with mathematical programming is proposed to address the computational complexity. This method models matches between couriers and order delivery legs via a hypergraph and drone movements via a time-space network. These components are integrated into a master trip assignment problem, which is solved iteratively using a column-and-row generation framework where new hyper-edges prescribing many-to-many courier-order matches are identified by solving smaller MILPs. Numerical experiments using real-world food delivery data demonstrate the solution quality and efficiency of the proposed method. The simulation results reveal that the coordinated courier-drone delivery services achieve about 8% improvements in both service rates and ground operational cost reduction compared to traditional door-to-door delivery models.