Comparison of automated vehicle struck-from-behind crash rates with national rates using naturalistic data
DOI: 10.1016/j.aap.2021.106056
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 performance of automated vehicles (AVs) by specifically comparing their struck-from-behind crash rates with national rates for human-driven vehicles. While previous research indicated that AVs experience a high proportion of rear-end crashes, no prior study had compared these specific crash types using equivalent definitions and data sources. The research was motivated by the significant public health and economic impact of rear-end collisions, which account for roughly one-third of all crashes in the United States. The primary objective was to determine if AVs are struck from behind at higher rates per distance traveled than conventional vehicles and to investigate potential causes, such as driving environment or vehicle behavior. The methodology utilized two primary datasets: crash records from the California Department of Motor Vehicles (DMV) for AVs and data from the Second Strategic Highway Research Program Naturalistic Driving Study (SHRP 2 NDS) for human-driven vehicles. The study analyzed 256 AV crash reports submitted between 2014 and 2020, focusing on incidents where the AV was in autonomous mode. To ensure comparability, the SHRP 2 NDS data was age-weighted to reflect the U.S. driving population, correcting for the study’s overrepresentation of high-risk age groups. Crash rates were calculated per million vehicle-miles traveled, with confidence intervals determined using a Poisson distribution. The analysis also categorized crashes by driving environment (urban, business/industrial, etc.) and vehicle state (moving vs. stopped) to isolate contributing factors. The results indicate that AVs are struck from behind at a rate of 17.2 crashes per million miles (95% CI: 14.2–20.7), which is significantly higher than the national average for human-driven vehicles of 3.6 crashes per million miles (95% CI: 3.0–4.3). However, this disparity nar considerably when accounting for driving environments. Human-driven vehicles in urban and business/industrial settings—environments where AVs like Cruise and Waymo predominantly test—experience struck-from-behind rates of 6.6 and 7.6 per million miles, respectively. When comparing AVs to human drivers in these specific environments, the differences in crash rates are not statistically significant. Furthermore, the study found that AVs are more likely to be struck when stopped (8.9 per million miles) than when moving (8.4 per million miles). In contrast, human-driven vehicles have much lower rates when stopped, particularly in urban areas. The significance of these findings lies in the identification of environmental and behavioral factors influencing AV safety. The data suggests that the higher overall crash rates for AVs are largely attributable to their testing in dense urban and commercial environments rather than inherent safety deficiencies. Additionally, the higher incidence of crashes while stopped implies that AV behavior regarding the timing and location of stops—such as at stop signs or during right turns on red—is a more plausible contributing factor than unexpected deceleration rates. The study concludes that while AVs experience more frequent minor rear-end collisions, these do not necessarily indicate dangerous performance, as the crashes are typically low-speed and result in minimal damage. The findings underscore the importance of context-specific safety assessments and transparent reporting for evaluating automated driving 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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | OpenAlex-citations | — | — | 1 | 2026-06-20 |
| archive | success | semantic_scholar | — | — | 6 | 2026-06-26 |
| extract | success | pdftotext | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-26 |
| chunk | success | chunk | — | — | 1 | 2026-06-26 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-26 |
| enrich | success | semantic_scholar | — | — | 4 | 2026-06-26 |
| promote | success | — | — | — | 1 | 2026-06-20 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-26 |
| verify | partial | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- incidence prevalence
- naturalistic crash near crash
- crash typology
- comparative international
- induced exposure
- urban rural setting
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