Real-time EEG artifact removal in simulated driving

Abstract

With the increasing demand for safer and better driving experience, driver assistance systems are receiving more and more attention in car industry. There are many data sources available to recognize driver’s brain state, such as brainwave (electroencephalography, EEG), image processing, and eye movement pattern analysis. EEG has several advantages over other data sources including directly and early detection of the brain state. However, EEG is very sensitive to artifacts such as eye blink and head movement. Traditional EEG artifact removal needs human experts to identify the artifact segment. In this FYP project, we aimed at removing EEG artifacts in real-time without human inspection. The objective of this project is to design scenarios that induce different kinds of EEG artifacts in a simulated driving, as well as an online real-time interface using machine learning algorithms to automatically recognize and remove artifacts. The project is interdisciplinary and includes research on driver’s safety, brain-computer interfaces, machine learning and signal processing. It is expected that the results of the FYP project could become a prototype for future car driver assistance systems. This interim report summarized the theory study and program implementation of Independent Component Analysis and Extreme Learning Machine algorithms for EEG artifact removal application. In addition, the report highlighted the progress so far and recognized the future project plan for the upcoming semester.