Models of human decision-making as tools for estimating and optimising impacts of vehicle automation
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of validating automated vehicles (AVs) through virtual testing, specifically focusing on the lack of mature models for human behavior in simulations. While virtual testing is essential for cost-effective AV validation, existing human road user models often lack the granularity to capture the perceptual and decision-making dynamics critical for human-AV interactions. The authors propose a novel modeling framework that conceptualizes human behavior as a network of interrelated perceptual decision processes, modeled using evidence accumulation mechanisms (drift-diffusion models). This approach aims to simulate situational awareness and complex, multi-stage decisions rather than just generating trajectories. The study develops and applies these models to two specific scenarios: (1) a pedestrian deciding whether to cross a street in front of an approaching AV, and (2) a human driver taking over control from an AV in a safety-critical rear-end situation. The pedestrian model utilizes interconnected decision units to assess visual looming (time-to-collision) and explicit signals of the vehicle’s intent to yield. The driver take-over model incorporates decision units for braking, lane changing, and gaze allocation, driven by visual looming and uncertainty resolution. Model parameters were manually tuned to reproduce qualitative behavioral patterns reported in prior empirical studies by Schneemann and Gohl (pedestrian interaction) and Gold et al. (take-over scenarios). The results demonstrate that the models successfully replicate observed human behaviors. In the pedestrian scenario, the model reproduced findings that some pedestrians wait for the vehicle to stop completely, while others cross based on perceived time margins without clear signals of intent. The simulations further revealed that AVs applying exaggerated deceleration or using external human-machine interfaces to communicate awareness significantly reduce interaction time loss by facilitating faster pedestrian decisions. In the take-over scenario, the model aligned with empirical data regarding maneuver frequencies (steering vs. braking) based on time-to-collision and initial situational awareness. The simulations suggested that take-over delays and reduced awareness of adjacent lanes significantly increase crash risk, particularly in less urgent scenarios. The significance of this work lies in demonstrating that detailed, cognition-based human models can be integrated into virtual testing to predict and optimize AV impacts on traffic flow and safety. The findings provide actionable insights for AV design, such as the benefit of exaggerated deceleration or explicit communication to pedestrians, and the critical need to enhance driver situational awareness during hand-overs to minimize collision risks. The authors conclude that while these are preliminary models requiring further validation, they offer a promising direction for simulating the fine-grained human factors necessary for comprehensive AV testing.
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 | 2026-06-07 |
| archive | success | canonical_url | — | — | 7 | 2026-06-09 |
| extract | success | cached | — | — | 2 | 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 |
| promote | success | — | — | — | 1 | 2026-06-07 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 8 | 2026-06-11 |
| verify | partial | — | — | — | 1 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- situational awareness
- mental model of traffic
- acceptance adoption
- ehmi external hmi
- automation surprise
- automation
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: tool software
- Theoretical Contribution: computational model, theory or model