Relaxation State Assessment on Sleepy Driver through EEG-Based Ensemble Learning

As the prevalence of sleepy driving continues to pose a significant threat to road safety, the need for alternative methods to assess the relaxation state of sleepy drivers becomes necessity. this study presents an alternative approach utilizing Electroencephalography (EEG) data and a synergistic ensemble learning on Neural Networks (NN) and Random Forest (RF) models with hyperparameter tuning. EEG signals, obtained from drivers in simulated or real-world scenarios, capture neurophysiological patterns indicative of relaxation levels. The proposed methodology involves meticulous preprocessing of EEG data to normalize features and split the dataset for training and testing. A Neural Network is designed to capture intricate patterns in the EEG data, while a Random Forest, optimized through hyperparameter tuning, enhances robustness and generalization. Ensemble learning, combining the strengths of both models, provides a comprehensive evaluation of the relaxation state. Results demonstrate the efficacy of the EEG-NN-RF ensemble in accurately assessing drivers’ relaxation states, paving the way for effective monitoring systems and adaptive driver assistance technologies. This research advances proactive measures to mitigate risks associated with sleepy driving, contributing to road safety and accident prevention.