A simulation‐based impact assessment of autonomous vehicles in urban networks

Sadid, Hashmatullah; Antoniou, Constantinos · 2024 · Crossref

DOI: 10.1049/itr2.12537

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 inconsistency in existing literature regarding the impact of autonomous vehicles (AVs) on urban traffic networks, which stems from reliance on assumed rather than optimized driving behaviors. Previous simulation-based assessments often modify car-following (CF) parameters based on limited field data or subjective assumptions, leading to conflicting conclusions about safety and efficiency. To resolve this, the authors propose a simulation-based impact assessment that utilizes optimized AV driving behavior derived from a particle swarm optimization (PSO) algorithm, rather than arbitrary parameter adjustments. The research focuses on a city-scale network in Munich, Germany, to evaluate how varying AV penetration rates (PRs) and demand scenarios affect traffic safety and efficiency. The methodology employs a microscopic traffic simulation framework using the AIMSUN Next platform. The base model, representing a fully human-driven vehicle (HDV) environment, was calibrated using PSO to accurately replicate real-world driving behaviors. For AVs, the study applied optimized CF parameters extracted from a prior optimization framework, ensuring the simulation reflects the most efficient and safe driving maneuvers possible for AVs under specific PRs. The experimental design involved running multiple simulation replications across different AV PRs and demand levels. Key performance indicators included the number of traffic conflicts for safety assessment and average network travel time for efficiency evaluation. Statistical analysis was conducted using generalized estimating equation (GEE) and zero-truncated Poisson (ZTP) regression models to quantify the impacts. The results indicate that AV deployment significantly improves traffic safety at higher penetration rates. Specifically, at a 100% AV PR, the total number of conflicts decreased by approximately 25% compared to a fully HDV environment. However, the study found that lower AV PRs could potentially lead to deteriorating safety conditions, likely due to the mixed-traffic interactions between cautious AVs and unpredictable HDVs. Regarding traffic efficiency, AVs did not demonstrate significant improvements; in some scenarios, they slightly increased average network travel times, although these changes were minimal. The findings suggest that while AVs enhance safety through optimized behavior and sensing capabilities, they do not necessarily alleviate congestion or reduce travel times in urban networks. The significance of this research lies in its methodological shift from assumption-based to optimization-based modeling of AV behavior, providing more reliable estimates of AV impacts. The study highlights that the benefits of AVs are not linear with respect to penetration rates, particularly concerning safety in mixed-traffic environments. These findings are crucial for policymakers and urban planners, suggesting that widespread adoption is necessary to realize safety benefits, while efficiency gains may be negligible. The work underscores the importance of using optimized behavioral models in simulation studies to avoid misleading conclusions about the potential of autonomous mobility systems.

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.

• traffic density