仿人足底肌电特征的机器人行走规划

Humanoid Walking Planning Based on EMG from Human Foot-bottom

模仿人类行走规律是规划双足机器人运动的基础.以往模仿人类步态主要通过视觉方法或惯性模块测量(Inertia measurement unit,IMU)方法捕捉人体特征点轨迹.这些方法不考虑零力矩点(Zero moment point,ZMP)的相似性.为解决该问题,本文提出了一种基于足底肌电信号(Electromyography,EMG)和惯性模块测量信号的混合运动规划方法.该方法通过测量足底肌电信号计算出足底压力中心的位置以及踝关节扭矩,结合惯性模块所测量的人体躯干和双足轨迹,来规划双足机器人的步态.首先,用肌电仪测量足底肌电信号,用惯性测量模块测量人体各肢体部分的姿态轨迹,经数据标定后作为仿人机器人的运动参考;然后,通过预观控制输出稳定的步态.为确保仿人行走的效果,基于人体相似性对运动数据进行了步态优化.实验验证和分析表明,EMG信号超前ZMP约160 ms,利用这个特性实现了对压力点位置的有效预测,提高了机器人在线模仿人类行走的稳定性.

The research on mimicking human-like walk lays a basis for biped walking motion. The conventiaonal method of realising human-like walk is using computer vision or inertia measurement unit （IMU） to capture the feature points of human body. However, these methods do not consider the zero moment point （ZMP） similarity. To tackle this problem, this paper proposes a hybrid motion planning method based on foot-bottom electromyography （EMG） signal and IMU measurement. This method uses the measurement of EMG under foot bottom to estimate the ZMP and ankle torque and plans the robot gait combining the measurement of trajectories of human trunk and feet. First, the bionic data from human is calibrated and transformed as motion reference for the robot. Then the dynamic balanced walking motion is generated by preview control. To ensure the human-like feature, the planned motion is optimised based on human similarity. Finally, it is validated by experiments and analysis that EMG signal is advanced 160 ms before ZMP change. The short time prediction of ZMP is realized with higher degree of likeness with human motion.