Driver behavior while using Level 2 vehicle automation: A hybrid naturalistic study

Cooper, JM; Crabtree, Kaedyn W.; McDonnell, Amy S.; May, Dominik; Strayer, Sean C.; Tsogtbaatar, Tushig; Cook, Danielle R.; Alexander, Parker A.; Sanbonmatsu, David M.; Strayer, David L. · 2023 · Cognitive Research: Principles and Implications

DOI: 10.1186/s41235-023-00527-5

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 driver behavior and safety implications associated with Level 2 vehicle automation (simultaneous Adaptive Cruise Control and Lane Keep Assist) using a hybrid naturalistic driving design. While previous research relied heavily on simulations, this work addresses the gap in real-world data by examining how drivers interact with automation over time, specifically regarding usage frequency, system warnings, driver arousal (fatigue and fidgeting), and secondary task engagement. The study was motivated by concerns that the shift from active control to passive monitoring may lead to vigilance decrements, increased distraction, or fatigue, though the generalizability of simulator-based findings to real-world commuting remains unclear. The researchers employed a 6–8 week naturalistic study involving 30 participants who drove commercially available Level 2 vehicles during their daily work commutes. A unique experimental control was implemented: one randomly selected day per week was designated as an “Automation: NO” day, where participants were instructed to drive manually, providing a baseline for comparison against days when automation use was permitted. Video cameras recorded driver behavior, which was coded using the BORIS software for automation state, system warnings, driving demand (weather, traffic, construction), fatigue signs, fidgeting, and secondary tasks (e.g., texting, radio listening). Data were analyzed using linear mixed-effects models to account for repeated measures and contextual variables. Key findings revealed that drivers used Level 2 automation over 70% of the time on interstates but were discerning in their application, opting for manual control under high driving demand conditions. Contrary to common safety concerns, automation use did not significantly increase driver fatigue or fidgeting compared to manual driving; rather, drivers tended to use automation when they were already at risk of fatigue due to low driving demand. The frequency of system warnings increased over the study period, suggesting that drivers adopted a more relaxed interaction strategy with practice. Regarding secondary tasks, engagement did not alarmingly change with automation; the only significant increase observed was in radio listening. Usage frequency and re-engagement times did not significantly change over the 8-week period, indicating stable behavior patterns rather than a progressive decline in vigilance. The study concludes that drivers exhibit a nuanced behavioral profile when using Level 2 automation, challenging assumptions that automation inherently leads to dangerous distraction or fatigue. Drivers appear to self-regulate, avoiding automation in complex conditions and not significantly increasing risky secondary tasks. The findings highlight the value of hybrid naturalistic designs that incorporate experimental controls to isolate the effects of automation from contextual driving factors. This research provides critical real-world evidence that Level 2 automation, when used by discerning drivers, does not necessarily compromise safety through increased inattention or fatigue, offering insights for future human-automation interaction guidelines and system design.

Key finding

In a 6-8 week hybrid naturalistic study (N=30) of five commercially available Level 2 vehicles, drivers engaged automation on >70% of permitted interstate driving, were less likely to engage automation under high driving demand, and showed increasing system-warning frequency with continued exposure; L2 use alone did not raise fatigue or fidgeting versus a within-subject manual control condition, and secondary task engagement did not increase alarmingly during automation.

Methodology

naturalistic

Sample size: N=30 (12 female, 18 male), age 18-55 (M=35.73, SD=9.34); vehicle assignment: 8 Tesla Model S, 6 Tesla Model 3, 1 Cadillac CT6, 6 Nissan Rogue, 9 Volvo XC90.

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 2026-06-10; verification: verified.

Topics

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

• temporal
• automation
• automation complacency bias
• automation surprise
• manual
• 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: observational prevalence, behavioral performance data
• Theoretical Contribution: conceptual framework