After Vehicle Automation Fails: Analysis of Driver Steering Behavior after a Sudden Deactivation of Control
DOI: 10.20485/jsaeijae.9.4_208
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 driver steering behavior during sudden take-overs after automation failure, specifically examining how initial vehicle states affect recovery safety. As SAE Level 2 and 3 automated systems require drivers to intervene when automation fails, understanding the consequences of the vehicle’s state at the moment of control transfer is critical for designing safe shared-control systems. The authors aim to validate a simplified two-point visual continuous control model of steering to assess whether it can predict the outcomes of emergency steering interventions based on initial conditions such as heading angle, lane deviation, and steering wheel angle. Data were collected from 44 participants driving in the National Advanced Driving Simulator (NADS-1). Participants engaged in a secondary number recall task designed to divert visual attention from the road. While distracted, the vehicle was forcibly drifted to create extreme initial conditions, including significant lane deviations and heading angles, without warning. Upon regaining control, drivers performed corrective steering maneuvers. The study analyzed the maximum lane deviation reached during recovery as a key safety metric. Additionally, the researchers applied a modified version of the Salvucci and Gray two-point visual control model, which incorporates perceptual cues (near and far point angles) and a parameter for neuromuscular lag, to simulate the drivers’ steering profiles. The results demonstrated that the initial heading angle, initial steering wheel angle, and initial lane deviation significantly influenced the maximum lane deviation during recovery. Linear regression models confirmed that these initial conditions, particularly the interaction between steering wheel position and heading angle, were strong predictors of recovery outcomes. When validating the steering model, simulations successfully replicated the participants’ heading angle and lane deviation trajectories. However, the model failed to accurately reproduce the detailed fluctuations and timing of the drivers’ actual steering wheel angle profiles. The model could approximate the overall path but did not capture the intermittent nature of human corrective steering actions. The study concludes that while the modified two-point visual control model is insufficient for replicating detailed steering dynamics, it serves as a useful design tool for assessing the consequences of vehicle state at the moment of automation failure. The findings highlight that providing drivers with favorable initial conditions—specifically regarding heading and steering angles—can reduce the risk of unsuccessful recoveries, such as collisions or run-off-road crashes. The authors suggest that while this linear continuous model is adequate for evaluating trajectory outcomes in extreme take-over scenarios, more complex non-linear models may be necessary to fully capture human steering strategies under severe conditions.
Key finding
The initial heading angle and steering wheel angle at the moment of automation failure strongly affect the maximum lane deviation during driver recovery, and a modified two-point visual control model can replicate vehicle trajectory outcomes but not detailed steering profiles.
Methodology
simulator
Sample size: 44
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-05 |
| archive | success | canonical_url | — | — | 1 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-07 |
| chunk | success | chunk | — | — | 1 | 2026-06-07 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-07 |
| enrich | success | semantic_scholar | — | — | 1 | 2026-06-06 |
| promote | success | — | — | — | 1 | 2026-06-05 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 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.
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
- Theoretical Contribution: computational model, conceptual framework