Real-Time Lane-Level Crash Detection on Freeways Using Sparse Telematics Data

TL;DR

This work tackles real-time crash detection with lane-level localization using sparse telematics data. It introduces an offline-online framework: offline calibration discretizes trajectories into lane-aligned cells to learn a vehicle intention distribution $P((x,y)\to(x',y'))$ and optimizes an alert threshold $T^*$ via F1-score; online detection maps new records to cells and computes three risk signals (position transition, speed deviation, and lateral maneuver) that accumulate on a cell-level risk map until $A_t(x,y)\ge T^*$ triggers a lane-level alert. On Wisconsin freeway data, the method achieves a crash identification rate of 75% with 0.6% non-crash misclassification, reaches 96% overall accuracy and an F1-score of 0.84, and detects 13% of crashes more than 3 minutes before the recorded time. The approach is lightweight, scalable, and does not require high-frequency or fixed-sensor data, making it practical for nationwide deployment using ubiquitous telematics to improve safety and reduce secondary crashes.

Abstract

Real-time traffic crash detection is critical in intelligent transportation systems because traditional crash notifications often suffer delays and lack specific, lane-level location information, which can lead to safety risks and economic losses. This paper proposes a real-time, lane-level crash detection approach for freeways that only leverages sparse telematics trajectory data. In the offline stage, the historical trajectories are discretized into spatial cells using vector cross-product techniques, and then used to estimate a vehicle intention distribution and select an alert threshold by maximizing the F1-score based on official crash reports. In the online stage, incoming telematics records are mapped to these cells and scored for three modules: transition anomalies, speed deviations, and lateral maneuver risks, with scores accumulated into a cell-specific risk map. When any cell's risk exceeds the alert threshold, the system issues a prompt warning. Relying solely on telematics data, this real-time and low-cost solution is evaluated on a Wisconsin dataset and validated against official crash reports, achieving a 75% crash identification rate with accurate lane-level localization, an overall accuracy of 96%, an F1-score of 0.84, and a non-crash-to-crash misclassification rate of only 0.6%, while also detecting 13% of crashes more than 3 minutes before the recorded crash time.

Figures (6)

• Figure 1: Overview of the real‐time crash detection framework
• Figure 2: Crash location and corresponding time-space diagram
• Figure 3: Telematics data coverage and density for the freeway segment
• Figure 4: Illustration of spatial–temporal discretization and vehicle intention distribution extraction.
• Figure 5: Sequential lane-level risk map figuress with annotated maximum risk scores.