Sober-Drive: A smartphone-assisted drowsy driving detection system

Abstract

Drowsy driving, a combination of sleepiness and driving, has become a worldwide problem that often leads to tragic accidents and outcomes. Existing research findings have shown that the percentage of closure of eyelid (a.k.a PERCLOS) is an effective indicator to evaluate the driver's drowsiness. We present the Sober-Drive system, which leverages PERCLOS for on-vehicle drowsy driving detection using smart phones. Specifically, Sober-Drive is built upon a number of indicators that are discrete-approximated from PERCLOS, blink time and blink rate, and it exploits the Neural Network to classify the eye “open/close” states. We developed a Sober-Drive prototype on Android smart phones, and we conducted extensive real-world experiments to evaluate its performance, the results of which show that Sober-Drive has a high detection rate of more than 90% for drowsy driving behaviors.