Driver Sleepiness Detection System Based on Eye Movements Variables

Abstract

Driver sleepiness is a hazard state, which can easily lead to traffic accidents. To detect driver sleepiness in real time, a novel driver sleepiness detection system using support vector machine (SVM) based on eye movements is proposed. Eye movements data are collected using SmartEye system in a driving simulator experiment. Characteristic parameters, which include blinking frequency, gaze direction, fixation time, and PERCLOS, are extracted based on the data using a statistical method. 13 sleepiness detection models including 12 specific models and 1 general model are developed based on SVM. Experimental results demonstrate that eye movements can be used to detect driver sleepiness in real time. The detecting accuracy of the specific models significantly exceeds the general model ( P < 0.001), suggesting that individual differences are an important consideration when building detection algorithms for different drivers.