Generalizing EEG-Based Classification for User-Independent Brain-Computer Interface

The emergence of brain-computer interface technology has changed the way of interacting between people and the computer. By applying this technology to individuals with impairments, it is possible to help them regain their mobility. Consequently, exoskeleton robots, guided by electroencephalograms (EEG), have been studied to provide assistance to these individuals. However, previous methods have struggled to achieve accurate classification of user intentions, often displaying an excessive sensitivity to input noise. Thus, there is a need to develop methods that are robust to noise and yield highly accurate results. In this research, I proposed a noise robust system for classifying user intentions based on EEG signals. The proposed system takes EEG signals as input and outputs commands that guide exoskeleton robots in assisting individuals with impairments. These commands encompass a range of fundamental movements, including running, forward and backward walking, maintaining a stationary position, and more. Through comprehensive experiments, the results obtained by the proposed method substantiate its superiority over prior approaches. I expect that this method holds the potential to significantly aid individuals in need, particularly those with impairments or undergoing rehabilitation.