一种基于肌电信号的自适应人机交互控制方法
An adaptive human-robot interaction control method based on electromyography signals
摘要点击 465  全文点击 57  投稿时间:2019-05-05  修订日期:2020-06-29
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2020.90316
  2020,37(12):2560-2570
中文关键词  踝关节康复机器人  肌电信号  阻抗控制  自适应控制  稳定性
英文关键词  ankle rehabilitation robot  EMG signals  impedance control  adaptive control  stability
基金项目  国家国家自然科学基金深圳联合基金项目(U1813214);辽宁省博士启动基金资助项目(20180540131)
作者单位E-mail
张弼 中国科学院沈阳自动化研究所 zhangbi@sia.cn 
姚杰 中国科学院沈阳自动化研究所  
赵新刚 中国科学院沈阳自动化研究所 zhaoxingang@sia.cn 
谈晓伟 中国科学院沈阳自动化研究所  
中文摘要
      针对踝关节康复机器人运动过程中的人机交互性问题, 本文提出一种基于肌电信号的鲁棒自适应人机交 互控制方法. 针对患者难以保持某一动作、肌电信号微弱等特点, 提出一种新的关节角度估计方法. 该方法充分利 用了踝关节运动时胫骨前肌与腓肠肌的拮抗关系, 将踝关节的动作类型与单个肌肉群的收缩进行关联, 利用归一化 的特征值完成运动意图的辨识和运动角度的估计. 为了保证人机交互的安全性, 提出一种刚度、阻尼参数在线自适 应调节的阻抗控制算法. 基于交互力矩对机器人末端的运动角度与运动速度实时进行调节, 使其对外表现出等效 柔性. 实验研究表明所提出的人机交互控制方法是有效的, 并具有一定应用前景.
英文摘要
      Aiming at the issues in human-robot interaction of ankle rehabilitation robot’s movement, this paper proposes a new surface electromyography (sEMG) signals based robust adaptive control strategy. Considering that sEMG signals of stroke patients are weak and it is difficult for them to maintain some certain actions, a new joint angle estimation method is proposed. The antagonistic relationship between the tibialis anterior muscle and gastrocnemius muscle during ankle joint movement is fully used in this method, and the motion type of the ankle joint is correlated with the contraction of a single muscle group. After the recognition of the motion intention and the estimation of the movement angle are completed by using the normalized characteristic value, the continuous and smooth angle estimation curve is obtained. To ensure human’s safety when the interaction torque suddenly increases, this paper designs an adaptive control law, in which the stiffness parameter and the damping parameter are adaptively tuned. By using the interactive torque, the moving angle and speed of the robot end are adjusted in real time, so that the external flexibilities have been achieved. The experimental results demonstrate that the proposed human-robot interaction control method is effective and has potentiol for practical application.