On-board Drowsiness Detection using EEG: Current Status and Future Prospects

Abstract

Drowsiness is a transition of psychophysiological state from alert towards sleep causing degradation in concentration, thereby increasing the response time. Drowsy driving is one of the leading causes of accidents in transportation sector. An on-board warning system which helps drivers with essential feedback about the onset of drowsiness by continuously monitoring divers' psychophysiological state can help to reduce the drowsy driving related accidents. Physiological signals are found to be most effective for continuous monitoring and better detection of drowsiness. Among all the frequently used physiological signals, Electroencephalogram (EEG), a record of the electrical activities of the brain, showed the strongest relation with drowsiness. Hence, EEG is widely considered as a reliable measure for drowsiness, fatigue, and performance evaluation. In this paper, EEG analysis for drowsiness studies, current findings and future directions of this field are briefly reviewed. Power spectral density (PSD) based features are found to be the most commonly used features for EEG based drowsiness studies. EEG low-frequency bands (delta, theta, and alpha), especially alpha band, shows an increase in band power during the drowsy state compared to alert state. In contrast, high-frequency bands (beta and gamma), specifically beta band shows a decrease in band power during drowsiness. In terms of brain regions, frontal, parietal, and occipital are suggestively informative, especially, alpha from occipital and beta from frontal are two potential indicators. Therefore, identifying informative brain regions with specific frequency bands will help to reduce the number of electrodes required to develop an effective EEG based drowsiness detection and warning system.