UAV Path Planning in Three-Dimensional Complex Environments

In the face of the complex three-dimensional (3D) flight environment under the constraints of terrain, no-fly zone, height limit, weather and other threats, the traditional reinforcement learning unmanned aerial vehicle (UAV) path planning algorithm has the problems of unreachable target points and dimension explosion. This paper proposes a path planning method for UAV in complex 3D environment. Based on the traditional reinforcement learning Q-Learning algorithm and the principle of artificial potential field (APF) algorithm, this method modifies the reward function generation mechanism in Q-Learning algorithm. The improved Q-Learning algorithm generates a dynamic reward function by judging the action of each step combined with environmental information. The reward function combines the good performance of the artificial potential field algorithm to make the reward accumulation process smoother. Finally, the typical complex mountain area scene is selected for simulation experiments. The experimental results show that the path planning algorithm designed in this paper can carry out feasible path planning for UAV in complex 3D environment.