A comparison of patterns and contributing factors of ADAS and ADS involved crashes
DOI: 10.1080/19439962.2023.2284175
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 investigates the patterns and contributing factors of crashes involving Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS), addressing a critical gap in traffic safety research. While ADAS and ADS adoption is rapidly increasing, existing literature has largely relied on datasets that either exclude ADAS crashes or focus solely on static pre-crash variables, ignoring the chronological progression of incidents. The authors argue that distinguishing between ADAS (SAE Level 2, requiring human engagement) and ADS (SAE Level 3+, sustained automation) is essential for developing effective safety countermeasures. To this end, the study utilizes the latest National Highway Traffic Safety Administration (NHTSA) crash reports, which uniquely contain data for both system types, to analyze how static factors and dynamic crash sequences jointly influence crash outcomes. The methodology involves a rigorous data screening process applied to 1,374 ADAS and 475 ADS crash reports recorded through November 2022. After removing duplicates, confidential records, and incomplete narratives, the final dataset comprised 92 valid ADAS and 100 valid ADS crash reports. The researchers extracted chronological event sequences from narrative descriptions using a modified encoding scheme. They calculated the dissimilarity between sequences using the Needleman-Wunsch algorithm and performed agglomerative hierarchical clustering to identify distinct crash patterns. To address missing data and class imbalance, categorical variables were aggregated, and missing values were imputed using the MissForest algorithm. Finally, logistic regression models were constructed to determine how static factors (e.g., automation level, speed limit) predict crash patterns, and how these patterns, combined with other variables, influence crash outcomes, specifically vehicle contact area and injury severity. The analysis identified five distinct crash pattern clusters: SV maneuvers (except stopping), CP rule violation, CP maneuvers, hit object, and SV stopping. The results indicate that automation level, speed limit, and vehicle speed are significant predictors of which crash pattern occurs. Furthermore, the crash pattern itself, along with incident time, roadway type, roadway surface, and vehicle model year, significantly associates with crash outcomes. Specifically, the study demonstrates that crash progression acts as an intermediate factor linking static environmental and vehicle conditions to final outcomes like contact area and injury severity. The findings highlight that ADAS and ADS crashes exhibit different characteristics, largely due to the varying completeness of reporting and the distinct operational roles of human drivers versus automated systems. The significance of this research lies in its comprehensive framework for modeling crash outcomes by integrating both static pre-crash variables and dynamic crash progression information. By distinguishing between ADAS and ADS, the study provides nuanced insights into the safety implications of different automation levels. The findings suggest that improving ADS/ADAS control algorithms and enhancing driver education are necessary to mitigate risks in mixed traffic environments. This work offers a methodological advancement for traffic safety research, demonstrating the value of sequence analysis in understanding complex crash dynamics, and provides empirical evidence to guide the design of testing scenarios and safety policies for automated vehicles.
Key finding
Automation level, speed limit, and vehicle speed are significant predictors of crash patterns, while crash patterns combined with environmental and vehicle factors determine crash outcomes such as contact area and injury severity.
Methodology
dataset
Sample size: 202
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 | — | — | — | 1 | 2026-05-28 |
| archive | success | canonical_url | — | — | 1 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-07 |
| chunk | success | chunk | — | — | 1 | 2026-06-07 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-07 |
| enrich | success | semantic_scholar | — | — | 4 | 2026-06-15 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | partial | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- naturalistic crash near crash
- pre crash contributing factors
- adas effectiveness
- crash typology
- incidence prevalence
- 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