An Efficient Approach for Driver Drowsiness Detection at Moderate Drowsiness Level Based on Electroencephalography...

Area of Science:

• Neuroscience
• Transportation Safety
• Biomedical Engineering

Background:

• Drowsy driving is a major global cause of severe road accidents.
• Effective detection methods are needed to mitigate risks associated with driver fatigue.

Purpose of the Study:

• To develop and evaluate a novel method for detecting driver drowsiness using electroencephalography (EEG) and vehicle dynamics data.
• To analyze EEG signals in relation to Observer Ratings of Drowsiness (ORD) for improved drowsiness assessment.

Main Methods:

• EEG data and vehicle dynamics (lateral position, steering angle) were collected from 19 healthy males in a driving simulator under alert and drowsy conditions.
• Five key EEG features (mean, standard deviation, kurtosis, energy, entropy) were selected using Neighborhood Component Analysis.
• Six classification algorithms were employed, including Classification Tree and Ensemble Regression, with a focus on single-channel P4 EEG data.

Main Results:

• Classification Tree and Ensemble Regression achieved over 87.5% accuracy in detecting moderate drowsiness using EEG data.
• Using only single-channel P4 EEG data yielded higher performance (up to 91.31%) compared to multi-channel EEG.
• The best drowsiness detection using vehicle dynamics data (KNN classifier) achieved 75.11% accuracy.

Conclusions:

• Driver drowsiness at a moderate level can be reliably detected using specific features extracted from single-channel P4 EEG data.
• This approach offers a promising, non-invasive method for real-time drowsiness monitoring in drivers.
• The findings highlight the potential of focused EEG analysis for enhancing road safety.