Driver Fatigue Prediction Using EEG for Autonomous Vehicle

In this paper, we propose a method to predict the eye state (opened or closed) by using measured electroencephalography signals which are processed using principal component analysis and support vector machine. These signals are used to predict the driver fatigue from his eye state so that during emergency the system would automatically change the driving mode form manual mode of driving to autonomous mode of driving. The signals from 14 electrodes are recorded using a commercial OpenBCI system along with a custom headset for recording the brain activity data and a video is recorded which is used for manually tagging eye state of the driver during simulated driving. This data was used to test the performance for classifying eye state using several types of support vector machines. The best performing classifier was found to be fine Gaussian support vector machine with a performance of 81.2% for classifying eye state and the system changes the driving mode of vehicle from manual to autonomous mode. The system can be further developed to track eye state for real time applications.