基于最优控制的仿人机器人行走振动抑制

Walking Vibration Suppression for Humanoid Robot Based on Optimal Control

  • 摘要: 针对仿人机器人行走过程中由腿部非刚性特性引起的振动,提出了一种基于最优控制的行走振动抑制方法.首先对振动进行建模,并将这一模型加入到原有机器人动力学模型中去.然后基于拓展的动力学模型,使用预观控制方法求取一条符合质心加速度约束的控制轨迹,作为求解最优控制问题的初始解.进而由此初始解出发,迭代求解此带约束的最优控制问题,利用得到的最优控制轨迹即能以前馈方式抑制行走过程中的振动.最后,使用预观控制方法和本文方法在仿人机器人“空”上进行行走对比实验.实验结果表明提出的方法显著减小了机器人行走过程中零力矩点(ZMP)的振荡和躯干的晃动.该方法对行走振动抑制的有效性得到了验证.

     

    Abstract: To deal with the vibration brought by the non-rigid characteristics of the leg mechanism during the humanoid robot walking, a walking vibration suppression method is proposed based on optimal control. Firstly, the vibration is modeled, and then the model is added to the original dynamics model of the robot. Then, the preview control method is used to generate a control trajectory subjected to the COM (center of mass) acceleration constraints based on the extended dynamics model, where the trajectory serves as the initial solution of the optimal control problem. And starting from this initial solution, the constrained optimal control problem is solved by iteration method, yielding an optimal control trajectory which suppresses the vibration in a feed forward manner. Finally, contrast walking experiments of the preview control method and the proposed method are performed on the humanoid robot Kong. The experimental results show that the proposed method significantly reduces the oscillation of the ZMP (zero momentum point) and the shaking of the torso during the walking process of the robot. The effectiveness of the method for walking vibration suppression is validated.

     

/

返回文章
返回