TScan–Stationary LiDAR for Traffic and Safety Applications: Vehicle Interpretation and Tracking

Tarko, Andrzej P; Romero, Mario A.; Bandaru, Vamsi Krishna; Lizarazo, Cristhian · 2021 · 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 details the development and implementation of TScan, a stationary LiDAR-based system designed for traffic monitoring and safety analysis at road intersections. Motivated by the need for accurate, long-term data collection on motorists and vulnerable road users to support autonomous vehicle integration and traffic safety improvements, the project builds upon prior feasibility studies (SPR-3831). The primary objective was to transition from research code to a deployable, real-world solution by constructing two trailer-based prototypes and developing associated engineering applications. The methodology involved significant hardware and software engineering. Two trailer-mounted prototypes were built, each featuring a detachable head unit containing multiple LiDAR sensors (Velodyne HDL-32E and Ouster OS1-64) to overcome range limitations and occlusion issues inherent in single-sensor setups. The software was rewritten in C++ from MATLAB to enable real-time processing, utilizing parallel processing and careful memory management. A critical technical challenge was integrating data from multiple sensors and units; the team developed algorithms to transform measurements into a common spatial and temporal reference frame, using GPS-based Pulse Per Second signals for synchronization. Additionally, because existing tools like the Surrogate Safety Assessment Model (SSAM) lacked support for diverse road user types (e.g., pedestrians), the team created a new file format (TSFF) and a custom traffic conflict detection technique. The findings demonstrate that the revised algorithms significantly improved tracking accuracy and reduced false positives in conflict detection. The system successfully tracks objects within a 200-foot range, estimating positions, speeds, and dimensions ten times per second, while classifying vehicles into cars, trucks, and two-wheelers. The prototypes were designed for ease of maintenance and security, with the sensor heads detachable for storage. The system generates approximately 35 gigabytes of data per hour per unit, which is processed locally and combined via post-processing modules. Engineering applications were developed to visualize trajectories, count traffic, analyze speeds, and identify pedestrian presence. The significance of this work lies in the creation of a scalable, operational tool for transportation agencies. By addressing the limitations of previous simulation-based analysis tools, TScan provides high-fidelity, real-world data for traffic conflict analysis. The project concludes with a plan to transfer the prototypes to the Indiana Department of Transportation (INDOT), including a collaborative training process to ensure end-user proficiency. This transition aims to facilitate continuous feedback and minor system modifications, ultimately enhancing the efficiency of traffic performance measurement and safety analysis in Indiana.

Key finding

The conversion of TScan algorithms to C++ and the implementation of GPS-based synchronization for multi-sensor units enabled real-time tracking and significantly reduced false positive traffic conflict detections.

Methodology

other

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