Defining Time-to-Collision Thresholds by the Type of Lead Vehicle in Non-Lane-Based Traffic Environments

Abstract

While anecdotal evidence suggests the consideration of a constant time-to-collision (TTC) threshold in advanced driver assistance systems, yet the prospect of the warning criterion acting as a catalyst for the evaluation of conflict severities in non-lane based traffic environments needs further assessment. The current study attempts to provide a copula-based framework to capture the potential dependence between TTC and centerline separation between the interacting vehicles in staggered-following events according to lead vehicle type. Detailed trajectory data extracted from the video footage of urban roads were used for the purpose of this study. Considering negative association between the considered variables and checked by the performance measures, the bivariate Frank copula model is found acceptable to model the bivariate dependency structure between TTC and CS. In particular, the results of the study corroborate a pragmatic decreasing relationship of TTC with increasing CS and decrease in lead vehicle size- a finding which indicates that the risk level of riders in staggered-following scenario varies by the type of front vehicle ahead and the lateral positioning of the vehicles. Based on the results presented in this paper, our recommendation is to use different minimum TTC thresholds for different centerline separations and leader-follower pairs to suitably evaluate traffic safety in such non-lane based traffic environments.