Automation Detection of Driver Fatigue Using Visual Behavior Variables

Yonggang, Wang; Jingfeng, Ma · 2018 · DOAJ

DOI: 10.2478/ace-2018-0023

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 driver fatigue, a major contributor to road traffic fatalities, particularly among commercial drivers engaged in long-duration transport. The research aims to establish the correlation between objective visual behavior variables and subjective fatigue levels to develop a real-time predictor for detecting fatigue and alerting drivers at critical moments. The methodology involved 36 professional drivers with normal vision and at least five years of experience, recruited from Jinan, China. Participants completed naturalistic driving tests on three distinct routes in Shandong, corresponding to continuous driving durations of 2, 3, and 4 hours. During these tests, eye-tracking technology (Smart Eye Pro 6.0) recorded ten visual behavior variables: four fixation metrics (pupil diameter, number of fixations, fixation duration, search angle deviation), three saccade metrics (number, speed, amplitude), and three blinking metrics (frequency, duration, closure duration). Subjective fatigue was quantified using the Stanford Sleepiness Scale (SSS). Data were analyzed using one-way ANOVA to assess variations across four age groups (<30, 30–40, 40–50, >50) relative to baseline values. Results indicated that driving duration significantly affected both visual behaviors and subjective fatigue. After 2 hours, only average closure duration and subjective fatigue levels increased by at least 20%. After 4 hours, all variables changed significantly except for the number of saccades and pupil diameter. Saccadic eye movements were identified as particularly sensitive indicators of fatigue. Age played a significant role, with older drivers (>50) exhibiting more pronounced changes in visual behaviors and higher susceptibility to fatigue compared to younger groups. For instance, the >50 group showed significant increases in fixation duration and decreases in saccade speed after 4 hours. The study developed a statistical predictor model for each age group, demonstrating high goodness-of-fit ($R^2$ values ranging from 0.858 to 0.995), which characterizes the relationship between visual behavior variations and fatigue levels. The findings confirm that visual behavior changes are positively correlated with pupil diameter, fixation duration, and blinking metrics, and negatively correlated with fixation count, search angle deviation, and saccade metrics. The study concludes that limiting continuous driving time and enforcing rest periods, especially for older drivers, are essential for crash prevention. The proposed predictor offers a viable tool for real-time fatigue assessment and alert systems. Limitations include the small sample size and potential inaccuracies in self-reported fatigue and environmental influences on eye movements.

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-prismaquant on 2026-06-26; verification: verified.

Topics

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

• drowsiness detection algorithms
• time on task
• drowsiness
• truck driver fatigue
• eye movements scanning
• 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, behavioral performance data
• Methodological Resource: tool software