Exploring the potential of market-available connected vehicle data in border crossing time estimation

Abstract

This study evaluates the potential of market-available crowdsourced driving data—connected vehicle (CV) data in estimating crossing times for passenger vehicles at ports of entry (POEs). Two months of CV data collected from a POE at the US-Mexico border in El Paso, Texas were processed using cloud computation tools to generate hourly aggregated border crossing times (CV-Time). In addition, this study also generated different variables to characterize the speed profile of CVs at different locations along a POE. Different regression models were developed to estimate border crossing times based on CV-generated variables and compared with ground truth observations from existing monitoring systems. The results show that the CV-Time is strongly correlated with the ground truth observations with a correlation rate of 0.82. The best-fitted Gradient Boost Regression model achieved an RMSE of 15.50 and MAPE of 25%. Our findings suggest that market-available CV data is promising for monitoring border crossing times, especially for supplementing physical monitoring systems when they are down for maintenance.