Exploratory analysis of injury severity under different levels of driving automation (SAE Level 2-5) using multi-source data
DOI: 10.21203/rs.3.rs-2824465/v1
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 limited understanding of injury severity outcomes in crashes involving vehicles with different levels of driving automation. While Advanced Driver Assistance Systems (ADAS, SAE Level 2) and Automated Driving Systems (ADS, SAE Levels 3–5) are widely deployed, research comparing their safety performance using real-world crash data remains scarce. The authors aim to identify factors influencing injury severity for both vehicle types to inform manufacturers, policymakers, and stakeholders regarding safe deployment and usage. To achieve this, the researchers constructed a multi-source dataset combining crash reports from the National Highway Traffic Safety Administration (NHTSA), the California Department of Motor Vehicles (CA DMV), and news outlets. After cleaning and deduplication, the final sample comprised 709 crashes involving ADAS-equipped vehicles and 571 crashes involving ADS-equipped vehicles. Data extraction from PDF reports and news sources utilized Large Language Models (LLMs) followed by manual verification. The study employed two separate random parameters multinomial logit models with heterogeneity in means and variances to analyze injury severity outcomes, categorized as no injury, minor injury, and moderate/severe injury. This modeling approach was selected to account for unobserved heterogeneity and was validated as statistically superior to fixed parameter models. The analysis revealed distinct operational differences between the two automation levels. Crashes involving ADAS vehicles predominantly occurred on highways (56%), whereas ADS crashes were more frequent in urban settings, specifically at intersections (45%) and on local streets (39%). ADS crashes also occurred under better driving conditions, with 94% happening in daylight and 90% on dry surfaces, compared to lower percentages for ADAS. Notably, 123 ADS crashes involved vehicles traveling the wrong way, a scenario absent in the ADAS dataset. Model results indicated that weather conditions, traffic incidents or work zones, vehicle sophistication (manufacture year and mileage), collision type, and impact location (rear or front) significantly influenced injury severity. For instance, lower mileage vehicles had different injury probability profiles, and wet surfaces interacted with mileage to affect outcomes. The findings provide an exploratory assessment of the safety performance of ADAS and ADS vehicles in real-world environments. By distinguishing the risk factors associated with each automation level, the study offers actionable insights for improving vehicle design and operational guidelines. The results suggest that safety strategies must be tailored to the specific operational design domains of each automation level, as ADS vehicles face different environmental and behavioral risks compared to ADAS vehicles. This work contributes to the broader goal of reducing road fatalities by guiding the development of safer autonomous 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 | Crossref | — | — | 1 | 2026-08-09 |
| archive | success | canonical_url | — | — | 1 | 2026-08-09 |
| extract | success | pdftotext | — | — | 4 | 2026-08-10 |
| clean | success | clean | — | — | 2 | 2026-08-10 |
| chunk | success | chunk | — | — | 2 | 2026-08-10 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 2 | 2026-08-10 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.6-27b-nvidia | summ-v5 | 2 | 2026-08-10 |
| tag | success | vector_similarity | — | — | 17 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-10 |
Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- adas effectiveness
- naturalistic crash near crash
- demographic disparities
- incidence prevalence
- driverless ads
- pre crash contributing factors
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
- Methodological Resource: dataset resource
- Theoretical Contribution: conceptual framework