Driving Fatigue Detection from EEG Using a Modified PCANet Method

Area of Science:

• Neuroscience
• Artificial Intelligence
• Transportation Safety

Background:

• Traffic accidents significantly increase with automotive industry growth.
• Driving fatigue is a primary contributor to a substantial portion of traffic accidents.
• Electroencephalography (EEG) offers a direct and effective method for detecting driver fatigue.

Purpose of the Study:

• To introduce a novel feature extraction strategy for EEG-based driving fatigue detection.
• To enhance classification accuracy and efficiency using a deep learning model.
• To address the dimensionality challenges of deep learning models in EEG analysis.

Main Methods:

• EEG signals were collected from six healthy volunteers during a simulated driving experiment.
• A modified PCANet model was developed, integrating principal component analysis (PCA) for dimensionality reduction.
• PCA preprocessing was employed to mitigate the 'dimension explosion' issue associated with PCANet.

Main Results:

• The modified PCANet method demonstrated high and robust performance in driving fatigue detection.
• Achieved a classification accuracy of up to 95%, surpassing conventional feature extraction techniques.
• Identified strong associations between driving fatigue and activity in the parietal and occipital brain lobes.

Conclusions:

• The study successfully applied a modified PCANet technique for EEG-based driving fatigue detection.
• This approach offers a feasible and highly accurate solution for real-time fatigue monitoring.
• Parietal and occipital lobe activity are key indicators of driving fatigue.