Study to Determine the Effectiveness of Deep Learning Classifiers for ECG Based Driver Fatigue Classification

Driving fatigue is one of the significant factor cause road accidents which often result in huge socio-economic loss to the country. The real-time, accurate driver fatigue and drowsiness detection can bring down the accident rate. This paper verify the superiority of the deep learning approaches over machine learning methods for driver fatigue detection using electrocardiography (ECG). Heart rate variability (HRV) derived from ECG, and directly controlled by the autonomic nervous system is a promising indicator for real-time driver fatigue estimation. The classification system use time domain, frequency domain and nonlinear HRV features to ensure high accuracy and detection rate. This study conducted on 10 healthy individuals in simulator driving environment. The deep learning architecture based on Stacked Autoencoders achieve an accuracy above 90% to determine the perceived fatigue in drivers.