Driver Drowsiness Detection Using Real-Time Heart Rate Variability Data

Driver drowsiness detection is considered a major reason for road accidents and crashes, resulting in fatalities and serious injuries. To determine the degree of driver drowsiness, researchers have examined techniques such as eye movement detection, facial recognition, movement detection, and electroencephalography (EEG) based systems. Although these methods are less useful in real-life applications, they typically carry limitations like high computational cost, intrusive hardware, and environmental sensitivity. The current study addresses the use of real-time heart rate variability (HRV) data, a trustworthy and non-intrusive physiological parameter, to recognise drivers who are drowsy while operating an automobile to try to solve these problems. For evaluating the HRV data and predicting driver drowsiness, an artificial neural network model is applied. Performance results show that the model is doing well; it has an accuracy of 91.36 %, an 83.33 % F1 score, 82.33 % recall, and 85.33 % precision. It also has an 80 % recall. The promising research outcomes face data imbalance issues and insufficient dataset size. Furthermore, incorporating additional features associated with physiological signals and driver behaviors could enhance the system's adaptability. Implementing the system into embedded hardware enables real-time monitoring and automatic warning notification. Research leading to efficient drowsiness detection systems has created opportunities for safer roads coupled with an accident rate reduction.