Detection and analysis: driver state with electrocardiogram (ECG)

Driver drowsiness, fatigue and inattentiveness are the major causes of road accidents, which lead to sudden death, injury, high fatalities and economic losses. Physiological signals provides information about the internal functioning of human body and thereby provides accurate, reliable and robust information on the driver’s state. In this work, we detect and analyse driver’s state by monitoring their physiological (ECG) information. ECG is a non-invasive signal that can read the heart rate and heart rate variability (HRV). Filters are applied on the ECG data and 13 statistically signifcant features are extracted. The
selected features are trained using three classifers namely: Support Vector Machine (SVM), K-nearest neighbour (KNN)
and Ensemble. The overall accuracy for two-classes such as: normal–drowsy, normal–visual inattention, normal–fatigue
and normal–cognitive inattention is 100%, 93.1%, 96.6% and 96.6% respectively. The result shows that two-class detection provides better accuracy among diferent states. However, the classifcation accuracy using Ensemble classifer came down to 58.3% for fve-class detection. In the future, better algorithms have to be developed for improving the accuracy of multiple class detection.