生体信号デバイスを用いた電動車いすの操縦に関する研究

―サポートベクターマシンを用いた脳波による意思判別―

抄録

In this paper, brain-machine-interface for electric wheelchair using a bio-signal headset Emotiv EPOC is proposed. An electroencephalogram (EEG) of concentrated state and of relaxed state is distinguished using support vector machine (SVM), and crunching is detected based on the threshold criteria for electromyogram (EMG). Then, the electric wheelchair is controlled so as to move forward when user's concentration on consciousness is identified and stop when user's crunching movement is detected. In addition, accuracy of the identification rate for EEG was investigated. As a result, the concentration on consciousness was distinguished with 77.5% accuracy within 3.0 seconds in wheelchair operation environment. Realization of left-turn and right-turn is left as future work.