A consequence-aware driving risk field framework with data-driven calibration for high-risk scenario identification.

Authors

Jun Liu, Huiqing Jin, Zhongxiang Feng, Zeyang Cheng, Linzhi Liu

Published in

Accident; analysis and prevention. Volume 236. Pages 108704. Aug 07, 2026. Epub Aug 07, 2026.

Abstract

Driving risk is a dynamically evolving process arising from the coupled effects of traffic interactions, the road environment, and driving behavior. Existing traffic risk assessment methods largely rely on crash outcomes or localized surrogate safety measures, making it difficult to obtain a unified risk representation that is consistent with crash consequences across different driving states. This paper proposes a driving risk field modeling framework that incorporates constraints imposed by crash consequences. By integrating scenario-level risk factors with vehicle-specific risk characteristics, the proposed approach provides an instantaneous risk score for a given driving state and supports the identification and ranking of high-severity crash scenarios conditional on crash occurrence. To improve the interpretability and internal consistency of the model, this paper develops a parameter calibration method using real-world crash data and microscopic traffic flow simulation data. The key parameters of the risk field are then systematically optimized via a differential evolution algorithm. Experimental results show that the constructed kinetic field captures both distance decay and velocity amplification effects. The composite driving risk metric exhibits a stable unimodal distribution on the logarithmic scale. After data-driven calibration, the proposed Driving Risk Field model improved regression-error-related metrics and showed competitive capability in identifying and ranking high-severity crash samples compared with the XGBoost baseline. Meanwhile, the differences in risk-field distributions across weather conditions and road types indicate that the model can reflect the influence of different scenario factors on risk scores at the internal response level. The findings provide a unified and interpretable modeling paradigm for crash-consequence-calibrated high-risk scenario scoring, supporting risk-scenario screening and traffic safety analysis.

PMID:
42566839
Bibliographic data and abstract were imported from PubMed on 08 Aug 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.