Abstract Manager

Video conflict analysis is increasingly used to diagnose and remedy safety issues. One of the challenges with video conflict analysis is the lack of benchmark values. Without any comparison point, it is difficult to determine the significance of the conflicts identified and to evaluate the safety performance of various sites.

To address this issue, an innovative methodology was developed using predictive statistical models based on conflict data. Predictive models based on collision data have been used in road safety for the past 30 years with the advent of Safety Performance Functions (SPFs). These predictive models are statistical models which relate independent variables, such as geometric characteristics (e.g., number of legs, presence of right turn channels), traffic operations (e.g., traffic control type, presence of red-light camera, traffic signal plans), and traffic exposure (e.g., AADT) with a dependent variable (number of collisions or conflicts).

This methodology was applied as part of a project completed for the Regional Municipality of York (‘York Region’). For this project, video conflict data was collected for twelve signalized four-leg intersections. The conflict predictive models were developed using functional forms of Generalized Linear Models with Negative Binomial or Poisson distributions. Models were developed for vehicle-pedestrian and vehicle-vehicle conflict types, as well as sub conflict types such as right/left turning motorists with pedestrians and right/left turning motorists with through motorists. These models were used to establish a theoretical baseline for the number of expected conflicts at a given location, which was then compared to the observed number of conflicts recorded during the video conflict assessment.

This presentation will describe the methodology developed to understand the significance of identified conflicts and to use observed conflicts to evaluate the safety performance of various sites. Practical examples will also be presented to show how the models were used to assist in identifying safety issues and remedial measures in York Region.