Drowsiness detection using Raspberry Pi for EVs and smart cars

Abstract

Drowsiness detection is highly significant in assuring the safety and effectiveness of intelligent automobiles and electric vehicles (EVs). It used to be that managing driver fatigue was only a question of comfort for contemporary transportation systems. However, with the rapid improvements that have been made in automotive technology and the growing prevalence of autonomous features, this need has developed into a fundamental requirement. Sleepiness detection systems perform the role of watchful co-pilots by continually monitoring the driver's behavior and sounding alerts or taking other appropriate actions when indicators of tiredness are identified. They are an effective strategy to limit the dangerous practice of sleepy driving, which is responsible for many motor vehicle accidents. These accidents are caused by a combination of factors, including fatigue, distraction, and inattention. In the current investigation, a Raspberry Pi is a real-time monitoring system to determine drowsiness. The dataset had one thousand unique images, each depicting a different feature of a real-world driving event. These images have been organized into the following four categories: open eyes (250 images), closed eyes (250 images), open mouth (250 images), and closed mouth (250 images). During this investigation, the experimental circumstances were looked at during daylight and the evening hours. For the system to function correctly, it relies on the Eye Aspect Ratio (EAR) algorithm and the facial landmarks method. The recommended strategy showed a higher degree of accuracy when put into practice. However, the study found that false negative blinks were noticed due to noise that could not be repaired within the collected signal. In the future, we want to concentrate our research efforts on determining whether or not the recommended technique is effective in a broader variety of contexts.