Advancing Crash Investigation With Connected and Automated Vehicle Data – Phase 2

Khattak, Asad J.; King, Meredith; Adeel, Muhammad; Usman, Sheikh M.; Mahdinia, Iman · 2023 · ROSA P / Collaborative Sciences Center for Road Safety

archive: archived pipeline: cataloged verified

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

Summary

This report, titled "Advancing Crash Investigation With Connected and Automated Vehicle Data – Phase 2," addresses the need to modernize crash investigation practices in response to the diffusion of connected and automated vehicle (CAV) technology. Current investigations rely heavily on event data recorder (EDR) data, which lacks critical information regarding vehicle trajectories, surrounding environments, and driver factors. The study aims to determine how data from automated vehicle (AV) sensors—such as LiDAR, cameras, and radar—can supplement traditional methods, identify gaps in AV safety performance, and assess law enforcement preparedness for utilizing this new data. The research employs a multipronged approach across three distinct components. First, the authors developed a data pipeline using CARLA simulation software to process raw sensor data from AVs. They analyzed 94 real-world AV crash cases from California, noting that 70% were rear-end collisions, and simulated two representative scenarios: a pedestrian-AV interaction and a rear-end conflict between an AV and a conventional vehicle. The pipeline integrated LiDAR point clouds, camera video, and inertial measurement unit data to reconstruct crash dynamics. Second, the study analyzed 260 AV collision reports from the California Department of Motor Vehicles. Using text analytics to extract variables from crash narratives and Bayesian analysis to handle small sample sizes, the researchers assessed how pre-crash conditions, AV driving modes, and crash types influenced injury and property damage severity. Third, the team surveyed law enforcement officers to evaluate their familiarity with AV technologies and current training curricula. Key findings indicate that AV sensors provide valuable trajectory and velocity data typically unavailable in EDRs. Simulation results demonstrated that LiDAR could detect incoming vehicles earlier than rear cameras and that conventional vehicles often failed to brake in rear-end scenarios. The statistical analysis of crash narratives revealed that AVs operating on ramps or slip lanes faced a 37.7% higher risk of occupant injury. Additionally, AVs in automated mode were more vulnerable to rear-end collisions, while injury odds were positively associated with intersections, recreational areas, and interactions with pedestrians or bicyclists. The law enforcement survey highlighted a significant need for standardized training on AV data retrieval and interpretation, leading to the development of a recommended list of training topics for investigators. The significance of this work lies in its demonstration that AV sensor data can enrich crash investigations by providing a comprehensive portrayal of crash events, including the state of surrounding objects and driver actions. The findings have implications for roadway design, suggesting that critical infrastructure like ramps and intersections requires better support for AVs. Furthermore, the study underscores the necessity for updated training protocols for law enforcement to effectively utilize emerging CAV data, thereby improving road safety and accident reconstruction accuracy in mixed traffic environments.

Key finding

Automated vehicle sensors provide valuable trajectory and velocity data for crash reconstruction, and AVs operating on ramps or slip lanes are associated with higher injury severity.

Methodology

mixed_methods

Sample size: 260

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. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

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.

• naturalistic crash near crash
• in depth crash investigation
• incidence prevalence
• causation analyses
• telematics crash prediction
• adas effectiveness

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