基于脑电α/β波的智能轮椅人机交互  

Intelligent wheelchair human-machine interaction using α/β wave of EEG

作　　者：张毅[1] 罗明伟[1] 罗元[2] 徐晓东[3] 

机构地区：[1]重庆邮电大学自动化学院,四川成都611731 [2]重庆邮电大学光电工程学院,重庆400065 [3]电子科技大学自动化学院,四川成都611731

出　　处：《华中科技大学学报（自然科学版）》2013年第7期109-114,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：科技部国际合作项目(2010DFA12160);国家自然科学基金资助项目(51075420)

摘　　要：针对现有的脑电控制智能轮椅的方法的不足,提出了一种基于脑电α/β波的智能轮椅人机交互方法.利用闭眼放松的脑电信号的α波控制智能轮椅前进,左右手运动想象脑电信号的β波控制智能轮椅左转和右转;同时,提出了一种带惩罚的RCSP特征提取算法.通过实验验证,得到该特征提取算法能使受试者控制智能轮椅的平均正确率,即三类脑电信号的在线识别率都大于85%,且最大正确率高达89.17%,接近传统方法对两类运动想象脑电信号的识别率.对受试者进行了脑电控制智能轮椅走一个'8'字形固定轨迹的实验,实验结果显示:每个受试者都能较好地完成实验任务,表明了该人机交互方法的可行性.On the basis of intelligent wheelchairs controlled by signals from EEG（electroencephalograph）,a human-machine interaction for intelligent wheelchair was presented based on the α/β wave form EEG.The intelligent wheelchair was controlled forward by the α wave of closing eyes EEG signals and was controlled left turn and right turn by the β wave left-right hands motor imagery EEG signals.Meanwhile,A RCSP（regularizing common spatial pattern） algorithm with penalty was also presented.It is verified experimentally that the average accuracies which the subjects control intelligent wheelchair,or the recognition rates of three kinds of EEG signals are more than 85%,and the maximum accuracy reaches as high as 89.17%,closing to the recognition rate by using the traditional methods for two kinds of motor imagery EEG signals.Finally,the experiment controlling intelligent wheelchair off a fixed trajectory with ＂8＂ glyph was operated by the subjects.The experiment results show that each subject can finish the experiment task,further showing that the interactive method is feasible.

关 键 词：脑电 智能轮椅 运动想象 特征提取算法 人机交互 

分 类 号：R318[医药卫生—生物医学工程]