Quadruped robot motion control method based on central pattern generator

The invention discloses a quadruped robot motion control method based on a central pattern generator. The method comprises the steps that S1, a gait planning network is constructed based on the central pattern generator of an SO (2) group and a radial basis function neural network; s2, introducing a simulation-based optimization algorithm to train parameters in the gait planning network; s3, constructing a motion planning network based on a reinforcement learning algorithm; and S4, performing motion control on the robot in combination with the output of the trained gait planning network and the output of the motion planning network. According to the method, efficient and stable movement of the quadruped robot is achieved, and the quadruped robot can generate periodic gaits and start to move only by switching states after parameter training. 本发明公开一种基于中枢模式发生器的四足机器人运动控制方法，所述方法包括：S1：基于SO(2)群的中枢模式发生器、径向基函数神经网络构建步态规划网络；S2：引入基于模仿的优化算法对步态规划网络中的参数进行训练；S3：构建基于强化学习算法的运动规划网络；S4：结合训练后的步态规划网络的输出和运动规划网络的输出对机器人进行运动控制。该方法实现四足