Guidelines for Evaluating Safety Using Traffic Encounters: Proactive Crash Estimation on Roadways with Conventional and Autonomous Vehicle Scenarios

Tarko, Andrew P.; Romero, Mario A.; Bandaru, Vamsi Krishna; Shi, Xueqian · 2023 · ROSA P / Purdue University. Joint Transportation Research Program

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 addresses the need for rapid, proactive road safety evaluation methods capable of monitoring the transition from human-driven to autonomous vehicles. Traditional safety management relies on reactive crash records averaged over several years, which is insufficient for the rapidly changing safety landscape. The study aims to resolve two primary hurdles in using traffic encounters (near-crash events) as safety surrogates: cumbersome field observations and the lack of proven methods for estimating crash frequencies from these events. The research focuses on developing and validating a system to observe traffic encounters and estimate annual crash frequencies using LiDAR-based data. The methodology utilizes the TScan system, consisting of trailer-mounted units equipped with dual LiDAR sensors and video cameras, deployed at selected intersections. The study improved the TScan’s object detection and tracking algorithms, specifically enhancing multiple LiDAR self-alignment, vehicle dimension and orientation estimation, and trajectory smoothing using a Kalman Filter model that includes jerk. An engineering application was developed to analyze vehicle trajectories, identify traffic encounters based on spatial, temporal, evasive action, and severity criteria, and visualize these events. A key modeling effort produced expansion factors to scale conflict-based crash estimates from short observation periods to annual frequencies. Field evaluations involved deploying TScan units at three intersections to collect trajectory data, which was compared against historical crash records. The improved algorithms successfully identified traffic encounters and conflicts, generating lists of potential events and estimated annual crash numbers. The study confirmed the method’s ability to estimate crash frequencies, building on previous validations for rear-end collisions, and expanded the approach to other collision types. The results demonstrated that the system could effectively detect conflicts and provide spatial visualizations similar to traditional collision diagrams, allowing for the identification of safety issues without waiting for actual crashes to occur. The significance of this work lies in providing a practical, efficient tool for proactive safety analysis. The report delivers comprehensive guidelines, user manuals for the TScan hardware, and engineering application manuals for processing data. By enabling the estimation of annual crashes from short-term observations of traffic conflicts, the method allows transportation agencies to monitor safety changes in real-time and identify countermeasures more rapidly than traditional reactive methods. This is particularly critical for managing the safety implications of emerging autonomous vehicle technologies. The upgraded TScan prototypes and associated documentation are available for implementation, supported by a collaborative training process with the Indiana Department of Transportation to facilitate end-user adoption.

Key finding

The TScan-based method for identifying traffic encounters and conflicts successfully estimates annual crash frequencies from short observation periods, with preliminary evaluations confirming its accuracy for rear-end collisions.

Methodology

field_study

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
• incidence prevalence
• crash typology
• induced exposure
• exposure measurement
• telematics crash prediction

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