Research on Route Optimization for Truck–Drone Delivery Considering En Route Synchronization

Truck–UAV collaborative delivery can improve last-mile logistics efficiency, but fixed-node rendezvous often causes waiting loss and service delay. To address this problem, this paper proposes a route optimization method integrating en route synchronization, pseudo-node insertion, and GAT-PPO. Pseudo-nodes are generated along truck travel arcs to provide flexible UAV recovery points, and a time-recursive simulation model is developed to evaluate makespan and total tardiness under soft time windows. In the proposed framework, GAT is used to capture spatial–temporal relationships among nodes, while PPO supports sequential routing decisions and UAV dispatch coordination. Experiments on Solomon VRPTW instances with clustered, random, and mixed customer distributions show that GAT-PPO achieves the shortest total travel distance, the lowest total tardiness, and the shortest completion time among Random, NN, NN+2-opt, MLP-PPO, ALNS, GA, and VNS. Ablation results further confirm the contributions of GAT, PPO, pseudo-node insertion, en route synchronization, and UAV collaboration. The results indicate that the proposed framework can effectively reduce synchronization waiting loss and improve the temporal efficiency of truck–UAV collaborative delivery.