Research on Multi UAV Path Planning Based on Deep Reinforcement Learning

Nowadays, the low cost and convenience of drone technology have made it widely used and have become a key research focus of strategic goals in various countries. Previous drone path planning methods have shortcomings such as low efficiency and slow speed. Although deep reinforcement learning has made many achievements in path planning for drones, there are still problems such as long training time and low planning accuracy. To address the aforementioned issues, we propose an improved dual delay depth deterministic policy gradient (TD3) algorithm (Improved Artificial Potential Field Method TD3: APF-TD3) for path planning of unmanned aerial vehicles. Firstly, this article adds the combined value of artificial potential field and virtual potential field to the state space of the dual delay depth deterministic gradient algorithm. In addition, this article proposes a new reward function to improve the problems of slow model convergence and inability to explore new states in reinforcement learning algorithms caused by sparse reward feedback from the environment. Finally, the proposed algorithm was validated in a simulation environment, and the results show that our method can generate collision-free and smooth motion trajectories in real time and improve motion planning performance significantly.