Gazing as actual parameter for drowsiness assessment in driving simulators
DOI: 10.11591/ijeecs.v13.i1.pp170-178
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, which contributes significantly to traffic accidents and fatalities. While various methods exist for detecting driver fatigue, such as vehicle trajectory sensors or bio-sensors requiring skin contact, these approaches often suffer from environmental limitations or driver discomfort. The authors propose using eye-gazing properties as a non-intrusive, quantitative parameter for assessing drowsiness. Specifically, the research investigates the relationship between eye-gazing metrics and drowsiness levels, aiming to identify features that can distinguish between alert, lightly drowsy, and heavily drowsy states to enable early accident prevention. The experimental design involved eleven healthy male participants who underwent eight trials each in a driving simulator. Subjects drove an automatic transmission vehicle on an obstacle-free oval track at a constant speed of 100 km/h for 50 minutes. Data collection utilized a head-mounted eye gaze tracker sampling at 30 Hz to record eye movement signals, while a web camera captured facial expressions for psychological assessment. Drowsiness was evaluated using Facial Expression Evaluation (FEE), a five-level scale (0–4) rated by four examiners every 30-second epoch. The researchers extracted nine features from the gazing signals, including the number of gazing frames, clusters, blink occurrences, and their respective ratios. An optimum threshold for defining "gazing" was determined individually for each trial based on the sum of differences in frame counts. Statistical analysis revealed significant differences in gazing properties across three drowsiness categories: alert (FEE = 0), lightly drowsy (FEE = 1–2), and heavily drowsy (FEE = 3–4). Features such as gazing frames, gazing clusters, and non-gazing clusters decreased significantly with increasing drowsiness, while blink frames and blink-related ratios increased significantly (p < 0.001 via Kruskal–Wallis test). To evaluate classification performance, the authors employed a Support Vector Machine (SVM) with recursive feature elimination. The SVM achieved an overall classification accuracy of 76.3% in distinguishing the three drowsiness categories. When simplified to a binary classification of alert versus drowsy (FEE = 1–4), the accuracy improved to 89.0%. The findings demonstrate that eye-gazing properties are effective parameters for quantitatively assessing driver drowsiness. The study confirms that specific gazing features, particularly those related to gaze stability and blink frequency, correlate strongly with subjective drowsiness levels. This approach offers a less intrusive alternative to electrode-based bio-sensors and overcomes the positioning limitations of camera-based systems. The results suggest that monitoring gazing signals can provide reliable, real-time detection of drowsiness, potentially enhancing driver assistance systems. However, the authors note that the study was conducted under monotonous driving conditions, and future work is needed to validate these features in more complex, real-world driving scenarios.
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | OpenAlex-citations | — | — | 1 | 2026-06-20 |
| archive | success | unpaywall | — | — | 2 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-20 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
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, validation psychometrics