基于强化学习的类人机器人步行参数训练算法

摘要

基于轨迹规划的类人机器人在合理的参数组合下可实现快速稳定的行走.为优化步行参数,提出一种基于强化学习的步行参数训练算法.对步行参数进行降阶处理,利用强化学习算法优化参数,并设置奖惩机制.在Robocup3D仿真平台上进行实验,结果证明了该算法的有效性.%Aiming at optimizing walking parameters for quick, and stable walking of humanoid robot based on trajectory planning method, this paper presents a walking parameters training algorithm based on reinforcement learning. By decreasing the number of walking parameters, the reinforcement learning is applied to optimize these parameters, and the reward and punishment mechanism is given. Experimental results show that the algorithm is feasible in the RoboCup3D simulation platform.