EEG-based Driving Fatigue Detection during Operating the Steering Wheel Data Section.

Area of Science:

• Neuroscience
• Transportation Safety
• Biomedical Engineering

Background:

• Detecting driving fatigue using Electroencephalography (EEG) signals is crucial for reducing traffic accidents.
• Current methods often exclude EEG data segments from steering wheel operation, potentially losing valuable information.
• Steering wheel operation is a common activity during actual driving, making its associated data relevant for fatigue analysis.

Purpose of the Study:

• To investigate the utility of EEG data segments recorded during steering wheel operation for driving fatigue detection.
• To compare the effectiveness of including steering wheel operation data versus excluding it.
• To evaluate the performance of logistic regression and multi-layer perceptron classification methods on this data.

Main Methods:

• Utilized spectral band power features calculated from EEG data, including segments from steering wheel operation.
• Performed cross-session and cross-subject verification using divided features.
• Compared logistic regression and multi-layer perceptron classification methods.

Main Results:

• The average accuracy difference between using data with and without steering wheel operation was only 2.27%.
• Multi-layer perceptron outperformed logistic regression by 10.37% in detecting driving fatigue.
• EEG data segments during steering wheel operation contain valid information for fatigue detection.

Conclusions:

• EEG data collected during steering wheel operation is a valid and valuable source of information for driving fatigue detection.
• Including this data does not significantly decrease accuracy and can enhance detection capabilities.
• Advanced classification methods like multi-layer perceptron show superior performance in analyzing this data for fatigue detection.