A model of dyadic merging interactions explains human drivers’ behavior from control inputs to decisions
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 paper addresses the challenge of modeling human driver interactions to improve the safety and social acceptability of automated vehicles (AVs). Existing driver models typically focus on isolated aspects of driving, such as gap acceptance or acceleration, and often rely on game-theoretic assumptions that treat drivers as rational, non-communicating agents. The authors argue that merging interactions are reciprocal and complex, involving high-level decisions (who merges first), joint safety margins, and low-level control inputs used for implicit communication. To bridge this gap, the study introduces a novel computational model based on the Communication-Enabled Interaction (CEI) framework. This model aims to capture all three levels of behavior simultaneously without assuming human rationality, explicitly accounting for how drivers communicate intent through vehicle kinematics and make decisions based on perceived risk. The researchers validated their model using empirical data from a coupled, top-down view driving simulator involving nine pairs of human participants. The experiment utilized a simplified merging scenario where two vehicles approached a single merge point with varying initial relative velocities and projected headways. Participants controlled acceleration to resolve conflicts while maintaining their initial velocity and avoiding collisions. The CEI model simulates drivers who maintain a deterministic plan for their own trajectory and form a probabilistic belief about the other driver’s future movements based on observed kinematics. If the perceived risk of collision exceeds a personalized threshold, the driver updates their plan. The model was fitted to the experimental data using grid search for individual risk thresholds and linear regression for incentive functions, simulating 990 trials to match the human dataset. The results demonstrate that the model accurately replicates human behavior across all three levels. Regarding control inputs, the model reproduced the characteristic piece-wise linear velocity patterns and the magnitude of acceleration/deceleration used by human drivers, including individual differences in how much each driver contributed to resolving the conflict. For safety margins, the model maintained joint gaps comparable to human drivers (mean difference of 0.3 m), capturing how individual contributions varied based on risk tolerance and incentive functions. Finally, the model correctly predicted high-level decisions regarding which driver merged first, replicating the statistical effects of projected headway and relative velocity on these outcomes. While minor discrepancies existed, such as the model slightly overestimating velocity effects and producing rare extreme braking events due to miscommunication, the overall qualitative and quantitative alignment was strong. The significance of this work lies in providing a unified framework that explains human merging behavior through communication and risk-based decision-making rather than rational utility maximization. By capturing the underlying mechanisms of interactive driving, this model offers a pathway toward developing interaction-aware AVs that can negotiate safely and legibly with human drivers. The authors suggest that this approach could generalize to other traffic scenarios, enhancing the fundamental understanding of human interactive capacities and improving the design of automation that adheres to natural traffic norms.
Key finding
A communication-enabled interaction model based on risk perception accurately predicts human drivers' control inputs, safety margins, and merging decisions in a simulated environment.
Methodology
simulation_modeling
Sample size: 18
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 author_sweep_intake on 2026-05-27.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-27 |
| archive | success | canonical_url | — | — | 11 | 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 | skipped | — | — | — | 4 | 2026-07-02 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| 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.
- gap acceptance
- mental model of traffic
- situational awareness
- driver vru interaction
- anticipation
- traffic density
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, theory or model