Key result
A transformer model combining frequency domain EEG features achieved 78% accuracy in detecting driver vigilance, outperforming previous dichotomous results.
Why the study?
Decreasing vigilance of safety drivers is a major cause of autonomous vehicle accidents, but little is known about the electrophysiological signals that assess their vigilance.
Does a transformer model using EEG signals improve driver vigilance detection accuracy?
Does a transformer model using EEG signals improve driver vigilance detection accuracy?
A transformer-based model using EEG signals achieves 78% accuracy in detecting driver vigilance, outperforming previous dichotomous methods.
May support EEG-based driver monitoring systems; leaves open prospective validation before clinical use.
Safety driver is one of the most important safety precautions to assure road safety in public road tests of autonomous vehicles (AVs), However, the decreasing vigilance of the safety driver has become a major cause of autonomous vehicle accidents in recent years. With the introduction of wireless, wearable Electroencephalography (EEG) devices and the enhanced computational capabilities of AVs, the ubiquitous monitoring of safety drivers as part of the emerging E-Health system has recently attracted great interest from researchers and industry. Unfortunately, we know little about the electrophysio-logical signals that assess safety driver vigilance. In this work, we propose a method for detecting driver vigilance based on EEG signals that combines frequency domain EEG features and a transformer model to maximize the prediction accuracy and recall rate. Experimental results show that trichotomous results based on the proposed model outperform previous dichotomous results, reaching 78% when tested with data from the Sustained-Attention Driving dataset published in Nature scientific data.
No takes yet. Share an insight, caveat, or question.
Gao et al. (2022) studied Decreasing driver vigilance. Transformer model combining frequency domain EEG features vs. Previous dichotomous models was evaluated on Prediction accuracy for driver vigilance. A transformer model combining frequency domain EEG features achieved 78% accuracy in detecting driver vigilance, outperforming previous dichotomous results.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: