Physiological Measures of Risk Perception in Highly Automated Driving

Perello-March, Jaume R.; Burns, Christopher G.; Birrell, Stewart A.; Woodman, Roger; Elliott, Mark T. · 2022 · Crossref

DOI: 10.1109/tits.2022.3146793

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 efficacy of physiological measures in detecting risk perception during Highly Automated Driving (HAD), specifically addressing the challenge of monitoring drivers who are out-of-the-loop. As HAD systems (SAE Levels 3 and 4) allow drivers to engage in non-driving related tasks, their situational awareness may diminish, complicating safe take-overs. The research aims to determine if cardiac and skin conductance indicators can reliably capture arousal variations associated with both long-term, low-to-moderate risk scenarios and short-term, high-risk events. The experiment utilized a high-fidelity driving simulator with 20 licensed drivers divided into two groups: one experiencing surrounding traffic and one without. Participants underwent an 11.5-minute trial comprising baseline rest, two minutes of suburban driving, two minutes of city driving with heavy rain and increased traffic density, and a sudden hazardous event involving a collision with a semitrailer. Physiological data were recorded using electrocardiogram (ECG) and electrodermal activity (EDA) sensors. Heart rate variability (HRV) metrics, specifically high-frequency power and RMSSD, were analyzed to assess vagal tone, while skin conductance responses (SCRs) were measured to detect sympathetic arousal. Statistical analyses included mixed ANOVAs to compare group differences and repeated-measures ANOVAs to evaluate changes across driving conditions. The results indicated that heart rate variability features were superior at capturing arousal variations during long-term, low-to-moderate risk scenarios. Specifically, HRV metrics showed significant changes in response to the gradual increase in traffic density and environmental complexity. In contrast, skin conductance responses were more sensitive to rapidly evolving situations associated with moderate-to-high risk, such as the sudden hazardous event. The presence of traffic did not produce significant group differences in baseline arousal, but the dynamic changes in driving conditions elicited distinct physiological patterns depending on the metric used. These findings suggest that future Driver State Monitoring (DSM) systems should adopt a multimodal approach, combining multiple physiological measures to effectively capture both long-term and short-term modulations of risk perception. Relying on a single metric may fail to detect critical changes in driver readiness. By integrating HRV for monitoring sustained situational awareness and SCRs for detecting acute threats, DSM systems can better assess driver availability and improve the safety of take-over requests in automated vehicles. This research highlights the necessity of moving beyond eye-tracking alone to include robust psychophysiological indicators for comprehensive driver monitoring.

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.

Topics

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

• situational awareness
• automation surprise
• stress driving
• workload measurement
• automation
• hands on hands off engagement

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
• Methodological Resource: validation psychometrics
• Theoretical Contribution: theory or model