Electrooculography signal as alternative method to operate wheelchair based on SVM classifier

People with impairment conditions have some difficulties with daily activities such as controlling a wheelchair. Biosignal is one of the promising methods as an alternative signal to build communication between humans and machines. In this study, an alternative method to control a wheelchair is proposed. A prototype wheelchair is controlled using an EOG signal based on an SVM classifier. Four types of gaze motions: right; left; up and down are clustered using the one versus one rule. The SVM is trained using 356 data and tested with 100 data. The trained SVM has an accuracy of 0.99 for all types of gaze motions. Precision of the up and down gaze motion is 0.96, while the right and left gaze motion has a precision of 1.00. The trained SVM is implemented for real-time control of the wheelchair prototype. Seven participants controlled the wheelchair and moved inside a trajectory from the start to the finish points. The result shows that participants could operate the wheelchair using the EOG signal, even though they had difficulties because of comfort and accuracy issues of the system.