基于便携式脑电信号采集器的脑-机器人交互系统  

Brain-robot interaction system based on portable brain signal collector

作　　者：支丹阳 杜秀兰[1] 赵靖[1] 吴正平 李伟[1,3] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]三江大学工程教育学院,南京210000 [3]加州州立大学贝克斯菲尔德分校计算机科学与电子工程系,贝克斯菲尔德93311

出　　处：《电子测量与仪器学报》2016年第5期694-701,共8页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(61473207)资助项目

摘　　要：针对现有高精度脑电信号采集设备体积笨重、成本高、无法普及的问题,开发了便携低成本的多通道脑电信号采集器,并以可编程可定制的乐高机器人为应用实例,搭建了集注意力训练、教育、娱乐于一体的脑-机器人交互系统。脑电信号采集器内置锂电池,体积为65 mm×40 mm×21 mm,重约60 g,以无线方式传输信号。在此基础上,结合典型相关分析(canonical correlation analysis,CCA)与快速傅里叶变换(fast Fourier transformation,FFT),提出了基于CCA-Wx-FFT的SSVEP特征提取算法,对采集器信号进行分析,以确保对机器人的精确控制。10名被试的测试结果表明,本系统获得了与高精度EEG设备相近的信号特征和控制准确率(92.6%vs 94.1%)。这一脑-机器人交互系统极大地降低了系统开发的成本,方便携带,易于扩展,对脑控的普及具有重要的应用价值。Most of the existing high-precision EEG acquisition systems lie in their volume and heaviness as well as high cost,which restrict their widespread applications. In this paper,a low-cost portable multichannel brain signal collector is designed and a brain-robot interaction system for attention training,education and entertainment is established by combining a programmable and customized Lego robot. The brain signal collector powered by lithium battery is only 65 mm × 40 mm × 21 mm in volume and 60 g in weight,and transmits signals wirelessly. In addition,this paper proposes the feature extraction algorithm CCA-Wx-FFT based on the canonical correlation analysis（ CCA） and the fast Fourier transformation（ FFT） for the analysis of steady-state visual evoked potential（ SSVEP） data and the precise control of the robot. The experimental results of ten subjects demonstrate that the developed brain signal collector delivers comparable signal characteristics and accuracy（ 92. 6% vs 94. 1%） with the high-precision EEG equipment. The brain-robot interaction system based on the proposed collector and the Lego robot has the advantages of low cost,convenient portability and easy expanding,so it is important to promote its popularization and practicability.

关 键 词：便携式 脑电信号采集器 脑-机器人交互系统 CCA权重系数 注意力 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TN911.7[自动化与计算机技术—计算机科学与技术]