Safety performance functions in a road environment with automated vehicles

Coropulis, Stefano; Berloco, Nicola; Intini, Paolo; Ranieri, Vittorio · 2025 · Crossref

DOI: 10.55329/fzyz2882

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 lack of quantifiable data regarding the safety impact of Automated Vehicles (AVs) on road networks. While AVs are considered a promising solution to reduce human-error-related crashes, current Safety Performance Functions (SPFs)—standard tools for predicting crash frequency—do not account for AV presence due to a scarcity of observed crash data. The authors propose a methodology to develop an ad hoc SPF specifically for multivehicle crashes in mixed traffic environments, incorporating AV market penetration rates to predict future safety outcomes. The research utilized microsimulation to generate crash data, as real-world AV crash datasets are insufficient. The study focused on 16 two-way, two-lane rural road sites in the Province of Bari, Italy. Using the Aimsun Next software, the authors simulated traffic scenarios based on Gipps car-following and lane-changing models. They defined distinct behavioral parameters for Regular Vehicles (RVs), Partially Automated Vehicles (PAVs, SAE levels 2–3), and Fully Automated Vehicles (FAVs, SAE levels 4–5), reflecting differences in reaction times, aggressiveness, and speed adherence. The simulations covered three time horizons (2030, 2040, 2050) with varying AV penetration rates derived from literature. To validate the model, simulated traffic volumes were compared against observed data using the GEH statistic, and simulated conflicts were converted to crashes using the Univariate Extreme Value approach based on Time To Collision (TTC) thresholds. The sensitivity analysis revealed that the coexistence of PAVs and FAVs reduced risky interactions compared to mixed scenarios with RVs. However, the simultaneous presence of FAVs and RVs increased rear-end and crossing conflicts, attributed to FAVs driving more closely and RVs misinterpreting driverless behaviors. The final SPF was developed by adding a coefficient for AV presence to the baseline equation, controlling for road geometry and traffic exposure. The fitted models demonstrated satisfactory goodness-of-fit metrics, including Cumulative Residuals plots. The study specifically targeted multivehicle crashes, excluding single-vehicle incidents due to the high uncertainty in simulating unpredictable events like sensor failures or cybersecurity attacks. The significance of this work lies in providing a practical, adaptable tool for practitioners to estimate road safety performance in the era of AV deployment without requiring large datasets of unknown variables. By integrating AV market penetration into traditional SPF structures, the model allows for the assessment of safety impacts across different technological adoption scenarios. The authors conclude that this approach fills a critical gap in traffic safety research, offering a framework that can be calibrated for other contexts using Crash Modification Factors, thereby supporting infrastructure planning and policy decisions during the transitional phase of AV integration.

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 2026-06-26; verification: verified.

Topics

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

• adas effectiveness

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: crash risk outcomes