Proactive safety analysis using roadside LiDAR based vehicle trajectory data

The underlying weaknesses of crash data have led to the shift of traffic safety analysis from reactive to proactive approaches. Using conflicts/near-crashes as crash surrogates is the most common technique to evaluate the safety levels of traffic facilities prior to the occurrence of crashes. However, the existing conflict-based proactive safety measures face several challenges such as complex data collection methods, selection of suitable surrogate indicators and their thresholds, reliability issues of crash-conflict relationships etc. The availability of advance sensor-based technologies has made it easier to capture road-user data at a microscopic level; however, the implementation of these data types towards traffic safety applications is very limited. In this regard, this study uses the emerging, cost-effective LiDAR sensor-based technology to develop a methodology for proactive safety analysis. High resolution microscopic trajectory data collected using infrastructure-based LiDAR sensors is used to identify traffic conflicts as crash surrogates. In addition, bivariate extreme value models are developed using proximity and evasive action based surrogate indicators to predict crash frequencies at signalized intersections. A case study was conducted to implement the developed methodology at 5 signalized intersections of Lubbock, Texas. The results indicate a correlation between the identified conflicts and the crashes, and further provide new insights into the crash risks at the intersections. Overall, the proposed methodology lays ground for using roadside lidar based trajectory data for proactive safety analysis of signalized intersections.