Multiple UAVs Path Planning Based on Deep Reinforcement Learning in Communication Denial Environment

In this paper, we propose a C51-Duel-IP (C51 Dueling DQN with Independent Policy) dynamic destination path-planning algorithm to solve the problem of autonomous navigation and avoidance multiple Unmanned Aerial Vehicles (UAVs) in communication denial environment. Our proposed expresses Q function output by network as distribution, which improves fitting ability value. We also extend single-step temporal differential (TD) N-step timing differential, solves inflexible updates differential. More importantly, use an independent policy achieve UAVs without any each other. case rejection, can consistency avoid greedy behavior UAVs. multiple-UAV scenarios, our work includes path planning, taking off from different initial positions, same position. The hardware-in-the-loop (HITL) experiment results show that is much more robust effective than original Dueling-IP DQN-IP algorithms urban simulation has similar effects shared but significant advantage running