Voluntary Eye Blink Detection Using Electrooculogram for Controlling Powered Wheelchairs Considering Environmental Information

Abstract

This paper describes the voluntary eye blink detection method using electrooculogram (EOG) for controlling a powered wheelchair. This study aims to apply double blink, left and right wink to control commands for a powered wheelchair such as go/stop, left and right turn. Our previous study showed that these eye blinks can be detected by using EOG and classification models specialized for individuals. Therefore, the persons who are not registered cannot use this system because the amplitude of EOG has an individual difference. In this study, we proposed the voluntary eye blink detection method that has robustness for individual difference of EOG amplitude by using correlation coefficient with template signal. As the result of simulations, an averaged accuracy of 98.05% was obtained, and the efficacy of powered wheelchair applied the proposed method was confirmed.