Joint Optimization of Energy Efficiency and Stable Communication Quality for Cargo UAV-Enabled Multi-Package Pickup and Delivery

In this paper, we consider a cellular-connected cargo unmanned aerial vehicle (UAV) system for multi-package pick-up and delivery within a designated area, in which the UAV departs from a warehouse carrying packages for delivery, completes these deliveries, collects packages for pick-up, and finally returns to the warehouse. The most critical aspect is maintaining a reliable communication link between the UAV and the ground base stations (GBSs) while accomplishing these logistics tasks. Specifically, our objective is to minimize the cargo UAV’s total energy consumption for pick-up and delivery while ensuring that the probability of the UAV’s trajectories being covered by GBSs during its mission remains above specified thresholds. To address the proposed problem, we introduce a two-phase hybrid genetic algorithm and deep reinforcement learning (HGADRL) framework to obtain the logistics service order of the cargo UAV as well as its flight trajectories. In the first place, genetic algorithm (GA) is employed to determine the service order for the UAV, and based on this order, the second phase utilizes deep reinforcement learning (DRL) to further optimize the flight trajectory of the UAV in each stage. To demonstrate the validity of the proposed approach, three benchmark schemes for comparison and the effect of the two-phase HGADRL framework under four different communication outage thresholds are conducted. Extensive experiments manifest that the proposed framework exhibits significant superiority in minimizing energy consumption for cargo UAV systems while maintaining a stable connection with GBSs.