When Is It Safe to Complete an Overtaking Maneuver? Modeling Drivers’ Decision to Return After Passing a Cyclist

Rasch, Alexander; Flannagan, Carol; Dozza, Marco · 2024 · OpenAlex-citations

DOI: 10.1109/tits.2024.3454768

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 addresses the safety challenges drivers face when completing overtaking maneuvers of cyclists, specifically focusing on the decision to return to the original lane. The primary motivation is to improve active safety systems, such as blind-spot detection and forward-collision warning systems, by modeling driver behavior to reduce false-positive alerts that erode user trust. The research aims to develop a probabilistic model that predicts when a driver initiates the return phase, balancing the risk of side-swiping the cyclist against the risk of a head-on collision with oncoming traffic. The researchers utilized two distinct datasets to ensure cumulative evidence: controlled test-track data involving 18 drivers and naturalistic driving data collected from rural roads in Sweden. They focused exclusively on "flying" maneuvers, where drivers overtake without significantly reducing speed. Using Bayesian discrete-time survival models, they analyzed time-varying inputs to predict the probability of return onset. Key explanatory variables included the longitudinal displacement of the ego vehicle relative to the cyclist, lateral distance to the cyclist, the presence of oncoming traffic, and the time-to-collision with that oncoming traffic. The models were evaluated using both in-sample and out-of-sample metrics to assess predictive accuracy and robustness. The results demonstrated that drivers primarily use the longitudinal displacement of the cyclist to time their return decision. Specifically, the probability of returning increases as the ego vehicle moves further ahead of the cyclist. Furthermore, the presence of an oncoming vehicle significantly accelerates the return decision, with drivers returning earlier as the time-to-collision with the oncoming vehicle decreases. The Bayesian models successfully captured these dynamics, showing that drivers balance the lateral clearance needed to avoid side-swiping with the temporal pressure imposed by oncoming traffic. The study found that the models performed well in predicting return onset probabilities, providing a reliable framework for understanding driver behavior in these complex interactions. The significance of this work lies in its application to intelligent transportation systems. By providing a probabilistic model of driver behavior, the study offers a method to tune active safety systems so that warnings are triggered within the driver’s "comfort zone," thereby enhancing acceptance and effectiveness. This approach allows systems to anticipate driver actions rather than reacting solely to kinematic threats, potentially reducing nuisance alerts. Additionally, the model supports the development of automated driving features that can mimic human-like decision-making, ensuring that autonomous vehicles complete overtaking maneuvers in a manner that is both safe and predictable to other road users.

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 qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• driver vru interaction

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).

• Theoretical Contribution: computational model