Study On The Decision-making Behavior Of Vehicles Getting Across Pedestrians

Mixed traffic flow, which is the main characteristic of metropolitans in China, is regarded as one of the primary reason for serious traffic congestion. For the signalized intersection, great interference occurs between the motorized vehicles and the pedestrians which are heading for different directions in the same phase. This interference severely reduces the traffic capacity of the intersection, deteriorate the hidden trouble in traffic safety and make the traffic management difficult to be implemented. Model focused on microcosmic traffic behavior is one of the powerful tools for the study on mixed traffic, regardless the lack of mature and sophisticated theory and technology.Combined with the project under subsidy of National Natural Science Foundation conducted by my supervisor "Intersection level of service evaluation studies based on simulation", this paper studies the microcosmic behavior of the interference between vehicles and pedestrians in signalized intersection. Using the interference between left-turn vehicles and counter bicycles for example, building a mathematical model focused on the interference of pedestrians to the vehicles. The main steps we perform the study are summarized as follows: firstly, collect real traffic flow data of two classical intersections in Beijing by means of taking videos; then, analyze the collected data and try to obtain the useful data and principles existing among them; use the refined data and the related traffic flow theory to build the model for left-turn vehicles, building a model based on binary logistic regression and classification tree respectively which can description the decision-making behavior when vehicles passing through pedestrians groups. Through comparative analysis, the model based on the two algorithms respectively can reflect the decision-making process when vehicles crossing pedestrians feasibility and authenticity.