An Intelligent Collision Avoidance Algorithm for Unmanned Surface Vehicles Based on Deep Reinforcement Learning

To address the obstacle avoidance challenge for unmanned surface vehicles, this paper presents a novel intelligent algorithm based on deep reinforcement learning. The algorithm incorporates human demonstration experience data for quick convergence and efficient decision-making. It features an end-to-end framework for multi-sensor data processing and immediate action decisions. Both simulation and deployment experiments evidence the superiority of this algorithm.