Spatio-temporal dynamic change mechanism analysis of traffic conflict risk based on trajectory data

Analyzing risk dynamic change mechanism under spatio-temporal effects can provide a better understanding of traffic risk, which helps reinforce the safety improvement. Traditionally, spatio-temporal studies based on crash data were mostly conducted to explore crash risk evolution mechanism from a macroscopic perspective. Dynamic change mechanism of short-term risk within a small-scale area deserves exploration, which cannot be captured in macroscopic crash-based studies. It is practical to analyze traffic conflict risk as a surrogate safety measure, which can preferably overcome the limitations of crash-based studies. This study aims to explore the spatio-temporal dynamic change mechanism of conflict risk based on trajectory data. Both conflict frequency and severity are integrated and assessed by applying fuzzy logic theory to develop the whole risk indicator. Tra-jectories on U.S. Highway101 from NGSIM dataset are utilized and aggregated. A two-step framework is pro-posed to analyze the risk dynamic change mechanism. The spatial Markov model is firstly applied to explore the transition probability of risk level, and then the panel regression approach is employed to quantify the rela-tionship between spatio-temporal risk and traffic characteristics. Modeling results show that (1) the dynamic change trend of safety states differs under different spatial lag conditions, and it can be well depicted by the spatial Markov model; (2) dynamic spatial panel data modeling method performs better than the model that only considers temporal or spatial dependency. The novel proposed framework promotes a systematic exploration of conflict risk from a mesoscopic perspective, which contributes to assess the real-time road safety more comprehensively.

Automated summary

This study aims to explore the spatio-temporal dynamic change mechanism of conflict risk based on trajectory data. By integrating and assessing conflict frequency and severity using fuzzy logic theory, the spatial Markov model and panel regression approach are employed to analyze the relationship between spatio-temporal risk and traffic characteristics. The modeling results show that the dynamic change trend of safety states differs under different spatial lag conditions, and the dynamic spatial panel data modeling method performs better than the model that only considers temporal or spatial dependency. This novel framework contributes to a more comprehensive assessment of real-time road safety from a mesoscopic perspective.