基于脑肌电信号的机械臂控制方法与实现  

Robotic Control Method and Realization Based on EEG and EMG Signals

作　　者：李想 乔志强 张忠海[1] 于功敬[1] 孙健[1] 成苈委 Li Xiang;Qiao Zhiqiang;Zhang Zhonghai;Yu Gongjing;Sun Jian;Cheng Liwei(Beijing Aerospace Measurement&Control Corp.,Ltd.,Beijing 100041,China;School of Automation,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]北京航天测控技术有限公司,北京100041 [2]北京邮电大学自动化学院,北京100876

出　　处：《计算机测量与控制》2019年第12期83-87,92,共6页Computer Measurement &Control

摘　　要：针对脑-机接口目前存在的输入信息源单一、特征识别准确率低、输出控制指令少的问题,提出一种基于脑肌电信号的机械臂控制系统;首先对单侧手臂肌电信号和左右手运动想象脑电信号进行同步采集,然后分别进行特征提取和分类识别;并最终将分类模型应用于机械臂的多指令实时控制中;实验结果表明:20名被试者均实现了机械臂的多指令实时控制,且各动作识别准确率平均达到了95%以上;该系统模型丰富了混合脑-机接口的多样性,为脑-机接口在机械臂的控制应用提供了理论依据和实践基础。Aiming at the problem that the current brain-computer interface has a single input information source,low feature recognition accuracy and few output control commands,this study presents a robotic arm control system based on EEG and EMG signals.Firstly,the unilateral arm myoelectric EMG and the left and right hand motion imaging EEG are acquired synchronously,and then feature extraction and classification recognition are performed respectively.Finally,the classification model is applied to the real-time control of the robot arm.The experimental results show that all the 20 subjects achieved real-time control of the manipulator,and the accuracy of each action recognition reached more than 85%.The system model enriches the human-computer interaction-mixed brain-computer interface diversity,and provides a theoretical basis and practical basis for the brain-computer interface technology for robotic control.

关 键 词：脑-机接口 特征提取 分类识别 机械臂 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]