Were they in the loop during automated driving? Links between visual attention and crash potential

Louw, Tyron; Madigan, Ruth; Carsten, Oliver; Merat, Natasha · 2017 · Crossref

DOI: 10.1136/injuryprev-2016-042155

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 investigates the relationship between drivers’ visual attention and crash potential when resuming manual control from partially automated driving (SAE Level 2). The research addresses the concern that prolonged automation can induce "out-of-the-loop" (OOTL) states, characterized by passive fatigue and task disengagement, which may impair a driver’s ability to detect hazards and avoid collisions. The authors aimed to determine how different levels of driver engagement during automation affect visual attention patterns upon system uncertainty and whether these patterns predict crash outcomes. The experiment utilized a driving simulator with 75 participants who underwent a between-participant design involving five OOTL manipulations: no manipulation, light fog, heavy fog, heavy fog combined with a visual quiz, and no fog combined with an auditory n-back task. Participants experienced six discrete car-following events, two of which were critical scenarios requiring manual intervention to avoid a collision. Eye-tracking data recorded gaze fixations, specifically analyzing the location and distribution of visual attention in the seconds following the removal of OOTL manipulations and the onset of an "uncertainty alert." Results indicated that OOTL manipulations influenced the initial point of gaze fixation after the manipulations ceased, with secondary tasks causing greater variance in fixation locations compared to fog conditions. However, these differences in visual attention resolved within one to two seconds, and the type of OOTL manipulation did not significantly affect crash rates. Crucially, the study found a distinct difference in fixation patterns between drivers who crashed and those who did not during the first critical event. Drivers who avoided crashes demonstrated a stable, gradual increase in fixations toward the road center. In contrast, drivers who crashed exhibited erratic eye movements, maintaining low fixation on the road center initially before a sudden, late surge in attention. This suggests that crash victims either succumbed to automation surprise or over-trusted the system, delaying their visual scanning until it was too late to respond effectively. The findings imply that automated driving systems must direct drivers’ attention to potential hazards at least six seconds in advance of an adverse outcome to allow for stable information acquisition and decision-making. The study concludes that while automation relieves moment-to-moment driving demands, ensuring driver safety requires systems that facilitate early and smooth re-engagement of visual attention. Furthermore, drivers must possess an accurate understanding of system capabilities to maintain appropriate situational awareness. These insights are critical for designing human-machine interfaces that mitigate the risks associated with transitioning from automated to manual control.

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 2026-06-09; verification: verified.

Topics

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

• looked but failed to see
• peripheral attention
• gaze based attention detection
• visual
• eye movements scanning
• situational awareness

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: behavioral performance data
• Methodological Resource: tool software
• Theoretical Contribution: conceptual framework