基于深度强化学习的无人艇控制研究

Research on Control of Unmanned Surface Vehicle Based on Deep Reinforcement Learning

论文以水面无人艇为切入点,利用深度强化学习技术,结合无人艇的数学模型,分析并设计环境的状态空间、动作空间和奖励.通过智能控制器与环境交互生成训练样本,训练网络以实现无人艇的运动控制.通过试验仿真验证,训练后的网络能够很好地对无人艇进行控制,相对于传统的PID控制算法在稳定性以及抗干扰能力上具有一定的优势.

Based on deep reinforcement learning technology,and mathematical model of the unmanned surface vehicle,its state space,action space,and reward of the environment is designed,and training samples are generated through the intelligent controller.Combined with the track control,the motion control of the unmamied surface vehicle is realized.It is verified through experimental simulation that the trained network can control the unmanned boat very well,and has a certain advantage over the traditional PID control algorithm in stability and anti-interference ability.