An EEG-based perceptual function integration network for application to drowsy driving

Chuang, Chun-Hsiang; Huang, Chih-Sheng; Ko, Li-Wei; Lin, Chin-Teng · 2015 · Crossref

DOI: 10.1016/j.knosys.2015.01.007

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the critical safety issue of drowsy driving by developing an EEG-based system to detect driver vigilance states. While electroencephalography (EEG) is a robust physiological indicator for fatigue, scalp recordings are often contaminated by artifacts and represent a mixture of brain and non-brain sources. Previous methods relying on single-channel or subject-dependent features have limitations in real-world applicability. This research proposes a "perceptual function integration network" that utilizes Independent Component Analysis (ICA) to extract multiple independent brain sources, thereby creating a subject-independent framework for classifying driving performance. The experimental design involved ten healthy volunteers participating in a 1.5-hour simulated driving task using a high-fidelity virtual reality environment with a motion platform. The simulation induced drowsiness through monotonous highway driving at 100 km/hr. An event-related lane-departure paradigm was used to measure reaction times (RT), which served as labels for four vigilance states: alertness, inattention, drowsiness, and abrupt-awaking. EEG data was recorded from 30 channels, preprocessed, and decomposed using ICA to isolate five brain sources of interest: frontal, central, somatomotor, parietal, and occipital. Spectral features were extracted via Fast Fourier Transform and dimensionality reduction techniques, including nonparametric weighted feature extraction. A multiple classifier system was constructed, where individual classifiers for each brain source were trained using algorithms such as Support Vector Machines (SVM), Gaussian Classifiers, and Radial Basis Function Neural Networks. Final decisions were made via majority voting. The results demonstrated that spectral power in the delta, theta, and alpha bands (1–12 Hz) significantly increased in drowsy states across all five brain regions, with beta band differences observed in central and somatomotor areas. The parietal source yielded the highest individual classification accuracy. The proposed integration network, which fused outputs from all five sources, achieved an overall classification accuracy of 88%. This performance was significantly higher than any single-source approach and outperformed models using raw spectral arrays without feature extraction. Nonparametric feature extraction combined with SVM classifiers proved most effective, particularly in handling small sample sizes and high-dimensional data. The study concludes that integrating information from multiple cortical sources provides a more robust and accurate assessment of driver vigilance than single-source methods. The findings highlight the involvement of frontal, central, somatomotor, parietal, and occipital regions in maintaining alertness, with drowsiness particularly impairing sensory integration and visual reception. The proposed subject-independent model offers a practical solution for real-time drowsy driving detection systems, though future work is needed to improve online artifact removal to enhance stability in noisy environments.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified_with_issues.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms
• drowsiness
• neuro workload indices
• microsleep
• vigilance

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data
• Methodological Resource: tool software