Detecting driver mental fatigue based on Electroencephalogram (EEG) signals during simulated driving

Abstract

More than 20% of road accidents are due to driver fatigue and considered to be major inference for transportation safety. This study aims to analyze fatigue state from the EEG signals of the drivers while performing a simulated driving task. Ten healthy participants involved in this study at three different timings. EEG signals were recorded continuously during monotonous driving in the simulator. Based on the video analysis the fatigue state and normal state related EEG signals were collected and decomposed using discrete wavelet transform. In this work the alpha and theta sub bands were considered for analysing the fatigue state of drivers. Features were extracted and classified using machine learning algorithm such as SVM, KNN and Ensemble. The result shows an accuracy of 97.8% in occipital region (O1 in alpha and O2 in theta) for classifying the fatigue state.