Considering both energy effectiveness and flight safety in UAV trajectory planning for intelligent logistics

Abstract

In low-altitude economic logistics scenarios, trajectory planning for unmanned aerial vehicles (UAVs) can be treated as a typical traveling salesman problem (TSP). High-rise buildings in urban areas not only severely impact the flight safety of UAVs, but also increase their energy consumption when avoiding obstacles, thereby affecting their delivery ranges. To address these issues, this paper proposes a two-stage trajectory planning solution called ACO-DQN-TP for logistics UAVs. In the first stage, the ant colony optimization (ACO) algorithm is applied to solve the sequence for multi-target point deliveries, to obtain the optimal flight paths. The ant tabu table is reopened to allow for retracing of the movement paths in order to avoid forward search dilemmas. In the second stage, a deep Q-network (DQN) is combined with the traditional artificial potential field method to enhance the interaction between UAVs and their environment. The rewards are accumulated using two potential functions generated based on the target points and obstacles, to minimize the changes in the yaw angles and smooth the flight trajectory of the UAV. Simulation experiments were conducted on UAV trajectory planning for delivery missions with four to ten target points. The simulation results show that the average path length obtained by ACO-DQN-TP is 65% and 79% shorter than that of Greedy+DQNPF and BACO, respectively, and the sum of turning angles along the path is 56% of Greedy DQNPF and 72% of BACO on average. It indicates that the proposed ACO-DQN-TP scheme not only optimizes delivery routes compared to traditional ACOs but also effectively controls the magnitude of the changes in heading angle during flight. This ensures flight safety for the UAV through obstacle avoidance while reducing flight energy consumption. In particular, the heading angle optimization mechanism proposed in this paper has universal guiding significance for low-altitude flights in the areas of traffic and transportation using UAVs.