Path Planning Algorithms for USVs via Deep Reinforcement Learning

With the application of Unmanned Surface Vehicles (USVs) in complex and changeable environments, therefore it is particularly significant to enhance its autonomous navigation ability. Deep reinforcement learning (DRL) is a emerging method that combines the perception ability of deep learning (DL) with the decision-making ability of reinforcement learning (RL). Since it was put forward, DRL has achieved remarkable results in theory and application. In addition, DRL has the ability of perception and decision-making simultaneously, and shows strong learning ability. Therefore, DRL algorithms can be applied to solve the path planning problem of USV. In this paper, pointing at solving the path planning problem of USV, the horizontal 3-DOF motion model of USV is established, and the DQN algorithm is applied to settle USV path planning problem. Finally, the above algorithm is verified on the simulation platform.