融合运动想象脑电与眼电信号的机械臂控制系统开发

脑机接口(brain computer interface, BCI)旨在通过脑电信号与外部设备通信，以实现对外部设备的控制。针对目前脑机接口系统中混合多种复杂生理电信号，并且输出控制指令较少的问题，本文提出融合运动想象(motor imagery, MI)脑电与眼电信号方法扩充控制指令的轻量级机械臂控制系统。该系统分阶段融合脑电和眼电信号两种生物信号，使用双次眼电作为任务开关，运动想象脑电信号控制机械臂运动，单次眼电控制阶段切换，实现了二分类运动想象生成多种控制指令，完成了对机械臂的连续控制。其中运动想象脑电信号使用提升小波变换(lifting wavelet transform, LWT)和共空间模式(common spatial pattern, CSP)结合的方法提取特征，并采用支持向量机(support vector machines, SVM)进行分类；眼电信号通过分析无意识眼电和有意识眼电的峰值来设置阈值进行区分。为了验证系统的可行性，设计了一项脑控机械臂自主服药实验，通过在线实验测试，被试通过使用脑电信号和眼电信号实现了机械臂控制，并完成了服药流程，有利于进一步推广脑机接口技术的实际应用。

Development of a robot arm control system using motor imagery electroencephalography and electrooculography

A brain computer interface (BCI) aims to control external devices by communicating with them via brain electroencephalography (EEG) signals. To tackle the issue of complex mixed multiple physiological electrical signals and low output control instructions in BCI systems, a lightweight robot arm control system that integrates motion imagery (MI) and electrooculography (EOG) signals was proposed to expand the control instructions. In this system, two biological EEG and EOG signals were integrated gradually. Robot arm movement was controlled by MI using double EOG as the task switch, and the control stages were switched by single EOG. The dichotomous MI generated a variety of control instructions, completing continuous control over the robot arm. MI EEG signals are extracted using lifting wavelet transform coupled with common spatial patterns and classified using support vector machines. The EOG signals were distinguished by analyzing the peak values of unconscious and conscious EOG set to a specific threshold. To verify the feasibility of this system, this study designed an autonomous drug-taking experiment. In the experiment, the subjects completed the drug-taking process using the BCI system with the robotic arm control, which is conducive to further promoting the practical application of BCI technology.