An Online Study to Explore Trust in Highly Automated Vehicle in Non-Critical Automated Driving Scenarios

Dong, Haoyu; Martens, Marieke; Pfleging, Bastian · 2021 · Crossref

DOI: 10.1145/3473682.3480259

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 gap in human-vehicle interaction research regarding trust calibration in non-critical automated driving scenarios (NADS). While existing literature predominantly focuses on take-over requests in safety-critical situations or full automation, this work investigates how situational trust fluctuates during ordinary driving events where the automated system handles dynamic tasks without requiring driver intervention. The authors argue that maintaining appropriate trust in NADS is crucial to prevent both over-trust, which may lead to misuse, and under-trust, which may cause disuse of available automation features. The primary research question explores how people’s situational trust changes continuously during these non-critical scenarios and identifies the factors influencing these changes. To answer this, the researchers conducted an online video-based experiment with 35 participants. Participants viewed five video clips depicting ordinary driving maneuvers—such as lane changes, overtaking, and navigating construction zones—from a driver’s perspective. Although the videos were originally recorded from manual driving vehicles, participants were instructed to imagine they were passengers in a highly automated vehicle. To measure trust continuously, the study employed a digital “Feeling of Trust” indicator located at the bottom right of the video player. This tool offered binary options: “Trusting” and “Not so Trusting.” Participants annotated their trust levels in real-time by hovering their cursor over the indicator when they felt distrust. After each video, semi-structured interviews were conducted where participants reviewed their annotations to explain the specific events and reasons behind their trust fluctuations. Data analysis involved mapping trust timelines to video events and coding interview transcripts to identify influential factors. The preliminary results demonstrate that trust is not static but fluctuates significantly even in non-critical scenarios. The study found both common patterns across participants and individual differences in trust responses. For instance, in one video clip, no participant maintained trust throughout the entire duration, whereas in another, nearly 40% of participants trusted the vehicle continuously. Specific events triggered consistent distrust, such as a vehicle entering a construction site with unclear road markings, poor visibility in tunnels, or aggressive merging behaviors by other cars. In one detailed case involving a merging orange car, participants lost trust when the ego-vehicle failed to adjust its speed or position adequately in response to the other car’s erratic maneuvers. Participants often compared the ego-vehicle’s behavior across similar events, noting inconsistencies that affected their trust. Subjective feedback highlighted concerns about the vehicle’s recognition capabilities, reaction timing, and overall maneuvering comfort. The significance of this work lies in its contribution to understanding the continuous nature of trust calibration in highly automated vehicles. By identifying that trust changes dynamically during routine driving, the study underscores the need for human-vehicle interfaces that support trust maintenance beyond critical take-over requests. The findings suggest that future designs should account for temporal and developmental perspectives of trust, potentially using adaptive interactions to explain vehicle behavior during ambiguous or complex non-critical events. This research provides a methodological foundation for further studies using simulators or Wizard-of-Oz setups to develop more nuanced trust models and interaction designs for automated driving systems.

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-nvidia on 2026-08-10; verification: verified_with_issues.

Topics

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

• trust calibration
• automation
• trust in automation foundations
• automation surprise
• acceptance adoption
• 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: self report data
• Theoretical Contribution: conceptual framework