Using driver monitoring to estimate readiness in automation: A conceptual model based on simulator experimental data

Gonçalves, Rafael C.; Goodridge, Courtney M.; Kuo, Jonny; Lenné, Mike G.; Merat, Natasha · 2024 · Crossref

DOI: 10.21203/rs.3.rs-4344023/v1

archive: archived pipeline: cataloged

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

Summary

This paper addresses the lack of standardized, objective methods for estimating driver readiness in SAE Level 2 automated vehicles. While Driver State Monitoring (DSM) systems are increasingly deployed to ensure drivers are prepared for takeover requests (TOR), there is no consensus on how to measure readiness or define safety thresholds. Existing approaches often rely on subjective self-assessments or complex machine learning models that lack interpretability, and experimental control of readiness is difficult due to individual variability. The study aims to provide a conceptual model and methodology for defining readiness thresholds using empirical data, without relying on subjective ground truth. The authors propose a conceptual model integrating evidence accumulation models (EAMs) with the concept of scenario controllability from ISO 26262. The methodology involves defining safety-critical scenarios based on exposure, severity, and controllability. The model uses EAMs to account for individual variability in how drivers accumulate cognitive and motoric resources over time. To validate this approach, the paper presents a proof of concept using previously collected experimental data from driving simulator studies involving SAE Level 2 automation. The analysis focuses on correlating objective behavioral metrics (such as gaze behavior and posture) with takeover performance, specifically crash avoidance probability, to establish readiness thresholds. The findings demonstrate that readiness can be objectively estimated by linking driver behavioral indicators to the probability of successful crash avoidance within a stipulated time budget. The model successfully maps the accumulation of evidence (representing resource availability) to scenario-specific controllability levels. By using crash avoidance as an objective ground truth rather than subjective ratings, the approach allows for the derivation of minimum readiness indicator values required for safe resumption of control. The study confirms that readiness is a dynamic variable that fluctuates with driver state and scenario demands, and that EAMs effectively capture the temporal aspect of resource accumulation necessary for safe takeover. The significance of this work lies in its contribution to the development of robust DSM systems. It provides a theoretical framework for translating experimental simulator data into actionable safety thresholds for vehicle manufacturers. By moving away from subjective assessments and opaque AI models, this methodology offers a transparent, evidence-based way to calibrate DSM warnings. This ensures that drivers are adequately prepared for critical transitions, addressing a key safety concern in the deployment of automated driving systems where drivers may be "out of the loop." The model supports the standardization of readiness estimation, facilitating safer human-automation interaction.

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.

• situational awareness
• automation
• acceptance adoption
• automation surprise
• mode awareness
• 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).

• Methodological Resource: validation psychometrics
• Theoretical Contribution: computational model, conceptual framework