Research On Pedestrian Crossing Behavior At Unsignalized Intersection Based On Trajectory Data

In recent years,the vehicle ownership and travel demand of urban residents in China have increased rapidly,which has led to the competition of road use rights between motor vehicles and pedestrians and other traditional road users at road junctions.Last year,the Ministry of Communications put forward the "car owner" policy and the pedestrian button measures,although the right to use the road can be clearly defined,but these measures still face great challenges.Direct and additional traffic management strategies can not always achieve the desired results.When the additional traffic management strategy fails,the problem of vehicle-pedestrian conflict in many environments will return to the purest situation of pedestrian road sharing,that is,the self-organizing form of vehicle-pedestrian without traffic lights.In this context,the study of pedestrian crossing decision-making mechanism is a fundamental issue in the analysis of pedestrian-vehicle conflict.Identifying and analyzing the influencing factors of pedestrian crossing decision-making is of great significance in explaining the process of pedestrian-vehicle conflict and constructing a reasonable pedestrian-vehicle interaction model.This paper studies the formation mechanism and behavioral characteristics of pedestrian crossing decision-making at unsignalized intersections,including three aspects:(1)Pedestrian trajectory parameter extractionUsing the "Traffic Trajectory Calibration and Analysis Software" developed by our research group,the trajectory data of pedestrians and motor vehicles are tracked,calibrated and collected under different traffic conditions at high and low peak,and the parameters covering pedestrian characteristics,vehicle characteristics and environmental characteristics are extracted,and the trajectory map of pedestrian movement is drawn for preliminary analysis.(2)Analysis of Pedestrian Crossing BehaviorA large number of behavioral characteristics parameters are extracted from the trajectory data to reveal the mechanism of pedestrian-vehicle conflict.By using the method of variance analysis,the behavioral characteristics of pedestrians,such as speed,waiting time,travel characteristics of companions and distance,are statistically analyzed,and the mechanism of the objective environmental factors on pedestrian crossing decision-making is revealed.(3)Modeling and analysis of pedestrian crossing decision-makingMultidimensional eigenvectors affecting pedestrian decision-making are constructed,and integrated learning algorithm is used for modeling and analysis.At the same time,unbalanced data processing method is introduced to improve the accuracy of the model.Finally,different factors affecting pedestrian crossing traffic flow decision-making are analyzed,and more accurate and abundant conclusions are drawn than previous studies.