The Stochastic Modeling of the Turning Decision by Left-Turning Vehicles at a Signalized Intersection in a University Campus

Abstract

The turning decision of a vehicle, at a signalized intersection, depends on the characteristics of the road users (e.g., vehicle, pedestrians, bicycles) and the intersection. The objective of this chapter is to estimate the turning decision of left-turning vehicles at a signalized intersection in a university campus. The signalized intersection, at the crossing of Boulevard de Maisonneuve Ouest and Rue MacKay within the Sir George Williams campus of the Concordia University (Montreal, Canada), is considered as a case study. The traffic video data were collected from 10 a.m. to 5 p.m. during the period of July–October in the year 2010. Vehicles turn at the intersection based on the gap between those and the crossing traffic, and complete the turning maneuver accepting the adequate gap (time or distance). The mean value of accepting the gap is known as the critical gap acceptance (CGA). The stochastic modeling of the left-turning decision is implemented at two stages—the estimation of the CGA by using probabilistic approaches; and the determination of the factors’ contribution by applying backpropagation neural network (BPN). The stochastic distribution functions estimate the CGA for passenger cars and other vehicles (e.g., buses, trucks, and vans) as 14.3 s and 16.5 s, respectively. The BPN models determine the bicycle distance from conflict point, platoon bicycles, existence of bicycle at conflict zone, bicycles’ speed, vehicles’ speed, pedestrians’ speed, number of vehicles passed, and vehicle moving at conflict zone are the predominant factors of left-turning decision.