Driver's Drowsiness Monitoring and Alarming Auto-System Based on EOG Signals

The main reason for road accidents is the driver's drowsiness which leads to a considerable number of car crashes, injuries, lots of fatalities, and significant economic losses. Driver's drowsiness is represented as a state which varies between sleep and wakefulness, that decreases cognitive skills and impacts the capability of performing the task of driving. This serious issue needs to develop an effective vigilance monitoring system capable of decreasing accidents by alerting the driver under various bad driving situations. For detecting drowsiness, vehicle-based methods (such as estimating the level of drowsiness depending on the movements of the steering wheel), behavioral-based methods (detecting the driver visual features using various resources such as facial expressions, eye movements, head movements, etc.), and physiologic-based methods (detecting the earlier stages of driver's drowsiness depending on physiological signals) can be utilized. This paper is focused on the designing and implementation of a driver assistance system which includes a driver's monitoring and alarming by using intrusive acquisition methods, called Electrooculography (EOG) signals. An embedded system based on ATmega2560 microcontroller on the Arduino board has been used to implement the EOG signal acquisition circuit. The developed system used several measurements to extract the features from EOG signals which makes it very sensitive to detect the driver's drowsiness. Furthermore, K Nearest Neighbors classifier (KNN) is used to give good accuracies. This system creates a low-cost device capable of quickly alerting the driver to ensure their safety. The experimental results show the efficiency and reliability of the proposed driver assistance system.