Assessing driver engagement in assisted driving: Insights from Pilot Evaluation, Focus Groups and driving simulator testing

Deiana, Francesco; Jackson, James; Periago, Cristina; Castro, Elena · 2024 · Crossref

DOI: 10.54941/ahfe1005781

archive: archived pipeline: cataloged

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

Summary

This paper presents a multi-phase methodology for assessing driver engagement in Level 2 (L2) automated driving assistance systems (ADAS), motivated by regulatory concerns from the European Commission, NHTSA, and EuroNCAP regarding driver attention and system limitations. The study, conducted by researchers at Applus IDIADA, integrates proving ground testing, focus groups, and a planned driving simulator phase to evaluate how drivers interact with medium versus advanced L2 systems. The primary objective is to develop a robust evaluation framework that combines subjective metrics (mental workload, trust) with objective performance measures to inform future safety regulations and automotive design. The first phase involved a proving ground test with 39 non-professional drivers (ages 21–58) comparing a 2020 Volkswagen Golf 8 (medium L2) and a 2020 Tesla Model 3 (advanced L2). Participants completed the Driving Style Questionnaire and rated their mental workload using the Integrated Workload Scale (IWS) and trust using the Trust in Automated System Survey (TASS) at five-minute intervals. Objective data included Time to Collision (TTC) during a critical event where an unexpected obstacle was placed in the lane, requiring a cut-out maneuver by the lead vehicle. The second phase consisted of three focus groups (24 participants total) held in late 2023, mixing "experts" from the track test with "novices" to gather qualitative insights. The third phase, scheduled for 2024 in China, will utilize a VI-grade dynamic simulator to incorporate physiological data (heart rate, respiration) and eye-tracking, adding a cross-cultural dimension to the study. Results from the proving ground test indicated that while both systems maintained low mental workload (no participant exceeded a 5/9 IWS score) and high trust (minimum average 80%), the advanced L2 system yielded significantly better engagement metrics. Participants in the advanced L2 vehicle reported lower mean mental workload (1.2–1.75) compared to the medium L2 vehicle (1.75–2.10) and higher mean trust (96.1% vs. 93%). Objectively, 60% of drivers in the advanced L2 vehicle achieved a TTC below 1.5 seconds, compared to only 45% in the medium L2 vehicle, suggesting the advanced system supports faster reaction times. Focus group analysis revealed a distinct generational gap: younger participants (20–45) expressed high trust and viewed ADAS as a cognitive enhancement, whereas older participants viewed systems as distracting or dangerous, often citing a fear of "unlearning" driving skills. Situational awareness emerged as a critical factor, with younger drivers valuing virtual vehicle representations that aid spatial understanding. Additionally, the "illusion of control" was identified, where high engagement led some drivers to mistakenly believe they were manually braking when the system was actually intervening. The significance of this work lies in its iterative, triangulated approach to validating driver engagement metrics, which is currently a key criterion in the EuroNCAP 2030 roadmap and Smart Cockpit assessments. By identifying specific differences between medium and advanced L2 systems and highlighting generational disparities in trust, the study provides actionable insights for designing interfaces that maintain appropriate driver alertness. The planned simulator phase aims to validate these findings in a controlled environment and explore cultural variations in driving habits, ultimately contributing to the development of safer ADAS implementations and supporting the transition toward higher levels of automation.

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 qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.

Topics

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

• automation
• hands on hands off engagement
• situational awareness
• automation surprise
• temporal
• trust calibration

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: self report data
• Methodological Resource: tool software, validation psychometrics