Study On Pedestrian Characteristics At Signalized Intersection

The urban traffic has the characteristics of mixing, low-speed and high density,which the mixed traffic flows of motor vehicles, non-motor vehicles and pedestrians are pervasive in. A large number of signals control technologies have been applied to the intersections to ensure the vehicles' operation order and pedestrians' safety at intersections. The pedestrians' traffic behavior characteristics have directly affected the efficiency of traffic intersections and the associated roads. Accordingly, it would be wise to analyze the pedestrians' behavior characteristics at signalized intersect-ions.Firstly, the tilt erecting camera is used to shoot the pedestrians' behavior at typical crossing signalized intersections, then convert the pedestrians' image plane coordinates to the world coordinates through the camera internal and external parameters calibration, and collect microscopic traffic parameters of pedestrian crossing such as pedestrians' displacement, walking speed, acceleration , trajectory.Secondly, the influence factors of pedestrians' crossing are analyzed by the pedestrians' velocity at the intersections in Chengdu,and establish a pedestrians'crossing behavioral decision model based on improved social force model. As the initial social force model can't simulate pedestrians' positions and changing speeds exactly, the relative velocity, desired speed and the force between pedestrians and obstacles are improved, then establish the decision model according to the three elements: pedestrians, vehicles, road. The parameters of the model are calibrated by the microscopic traffic parameters.Finally, the conflicts between pedestrians and turning vehicles are analyzed based on critical gap theory. As the turning motor vehicles have the greatest impact on pedestrians' decision-making at intersections without pedestrian crossing phase,calculated pedestrians' critical gaps of conflicted points by the Raff method based on the data of pedestrians' acceptable gaps and maximum rejection gaps, and the results are verified by the maximum likelihood estimation algorithm.