Safety of Mixed Traffic

Huang, Chunxi; Li, Tianhao; He, Dengbo · 2025 · Sharing the Road

DOI: 10.1201/9781032674667-5

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This chapter reviews empirical studies on the safety of mixed traffic, specifically focusing on interactions between automated vehicles (AVs) and human-driven vehicles (HDVs). The authors motivate this review by highlighting the limitations of simulation-based approaches, which often rely on unrealistic assumptions regarding vehicle behavior, communication perfection, and driver adaptation. To provide more realistic insights, the chapter synthesizes findings from reactive safety analyses (crash data) and proactive safety analyses (surrogate measures and behavioral studies). The review examines crash analysis methodologies, noting a shift from descriptive statistics in early studies to advanced techniques such as logistic regression, machine learning (e.g., XGBoost, decision trees), text analytics, and Bayesian models as data availability increased. A primary finding concerns the prevalence of mixed traffic crashes. Comparisons of AV crash rates against conventional vehicles yield mixed results due to differences in data sources and adjustments for underreporting. However, when controlling for severity and reporting biases, some studies indicate AVs have lower rates of severe crashes but higher rates of minor incidents. Crucially, rear-end collisions dominate AV-involved crashes, accounting for over 50% of incidents in several datasets. These crashes frequently involve AVs being struck from behind by human drivers, particularly when AVs are stopped or moving slowly at intersections. This pattern is attributed to AVs’ conservative driving styles and human drivers’ inability to anticipate AV behaviors or maintain safe following distances. Further analysis identifies specific factors associated with crash occurrence and severity. Environment-related factors include intersections, crosswalks, poor lighting, wet pavements, and complex road geometries. Motion-related factors involve AV maneuvers such as turning, slowing down, and stopping. Regarding crash outcomes, rear-end collisions, multi-vehicle collisions, and instances where the AV is at fault are associated with higher injury severity. Environmental conditions such as extreme weather, high-speed locations, and dark lighting also significantly increase the likelihood of physical injury. The chapter begins to outline proactive safety analysis by identifying key mixed traffic scenarios, including car-following, lane-changing, and car-passing, which are critical for understanding longitudinal and lateral vehicle interactions. The significance of this review lies in its synthesis of empirical evidence that challenges purely simulation-based conclusions. It highlights that current AV safety risks are largely driven by human driver errors and the mismatch between human expectations and AV conservative behaviors. The findings provide a foundation for designing targeted countermeasures, such as improving AV predictability and addressing specific high-risk scenarios like intersection stops. The authors emphasize the need for future research to account for biases in crash rate estimation and to further explore behavioral adaptations in mixed traffic environments.

Key finding

Rear-end collisions dominate AV-involved crashes, primarily occurring when human-driven vehicles strike automated vehicles from behind at intersections due to a mismatch in driving expectations and conservative AV behaviors.

Methodology

review

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 2026-06-10; verification: verified.

Topics

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

• incidence prevalence
• naturalistic crash near crash
• crash typology
• pre crash contributing factors
• adas effectiveness
• motorcycle crash typology

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, observational prevalence
• Methodological Resource: dataset resource