EEG signal analysis for the assessment and quantification of driver’s fatigue

Abstract

Fatigue in human drivers is a serious cause of road accidents. Hence, it is important to devise methods to detect and quantify the fatigue. This paper presents a method based on a class of entropy measures on the recorded Electroencephalogram (EEG) signals of human subjects for relative quantification of fatigue during driving. These entropy values have been evaluated in the wavelet domain and have been validated using standard subjective measures. Five types of entropies i.e. Shannon’s entropy, Rényi entropy of order 2 and 3, Tsallis wavelet entropy and Generalized Escort-Tsallis entropy, have been considered as possible indicators of fatigue. These entropies along with alpha band relative energy and (α + β)/δ1 relative energy ratio have been used to develop a method for estimation of unknown fatigue level. Experiments have been designed to test the subjects under simulated driving and actual driving. The EEG signals have been recorded along with subjective assessment of their fatigue levels through standard questionnaire during these experiments. The signal analysis steps involve preprocessing, artifact removal, entropy calculation and validation against the subjective assessment. The results show definite patterns of these entropies during different stages of fatigue.