Measuring and Predicting Drivers’ Takeover Readiness and Supporting Takeover Transitions in Automated Driving
archive: archived pipeline: cataloged verified
Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)
Summary
This technical report addresses the safety challenges associated with Level 3 automated driving, specifically focusing on the transition of control from the vehicle to the human driver. As automation allows drivers to engage in non-driving related tasks, they often become "out of the loop," making it difficult to resume control safely when the system reaches its limits. The study aims to facilitate these takeover transitions by (1) predicting driver readiness using physiological data and machine learning, and (2) designing support systems to guide driver attention during takeover events. Part 1 of the study employed a driving simulator experiment with 32 participants who performed a secondary task (playing Tetris) while the vehicle operated in automated mode. Researchers collected physiological data, including heart rate, galvanic skin response (GSR), and eye-tracking metrics, prior to takeover requests. To quantify takeover performance, the authors introduced a novel metric based on Fréchet Distance, which measures the similarity between the driver’s actual trajectory and a theoretically optimal trajectory, capturing both temporal and spatial quality. Five machine learning models were tested to predict this performance using pre-takeover physiological windows ranging from 1 to 20 seconds. The results indicated that Random Forest models significantly outperformed linear models and decision trees, capturing approximately 70–94% of the variability in takeover performance. The optimal prediction window was identified as 11 seconds prior to the takeover request, with high performance sustained between 9 and 14 seconds. Feature importance analysis revealed that GSR indices were the strongest predictors of readiness, followed by heart rate indices. Additionally, personalized models showed potential for increased accuracy compared to generalized models. Part 2 focused on supporting driver attention by identifying hazards that co-occur with takeover requests using naturalistic driving data from the Integrated Vehicle-Based Safety System program. Based on these findings and the N-SEEV model of visual attention, the researchers designed and evaluated a gaze guidance system in a simulator study. Drivers were exposed to takeover scenarios with either high-salience or low-salience attention guidance. The study found that drivers using the highly salient guidance system were significantly less likely to collide with secondary hazards during the takeover transition. These findings suggest that combining physiological-based readiness prediction with active attentional support systems can enhance safety during automated driving transitions.
Key finding
Random forest machine learning models utilizing galvanic skin response and heart rate data from an 11-second pre-takeover window most accurately predicted driver takeover performance, while highly salient gaze guidance systems reduced collision risks with secondary hazards during takeover transitions.
Methodology
simulator
Sample size: 32
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. Discovered via bulk_ingest_aaa_foundation on 2026-05-23 (6 acquisition events logged).
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | aaa_foundation | — | — | 2 | 2026-05-23 |
| archive | success | — | — | — | 1 | 2026-05-23 |
| extract | success | cached | — | — | 2 | 2026-06-10 |
| clean | success | — | — | — | 1 | 2026-06-01 |
| chunk | success | — | — | — | 1 | 2026-06-01 |
| embed | success | — | — | — | 1 | 2026-06-02 |
| enrich | success | — | — | — | 1 | 2026-05-23 |
| promote | success | — | — | — | 1 | 2026-05-23 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 3 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 19 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
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.
- takeover transitions
- automation
- situational awareness
- automation surprise
- temporal
- drowsiness detection algorithms
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: behavioral performance data
- Methodological Resource: measurement protocol
- Theoretical Contribution: conceptual framework