Rider's Heterogeneity-based Microscopic Modeling And Simulation Of Non-motorized Flow In Complex Traffic Environment

The growing popularity of electric bicycles as well as the bike-sharing scheme launched in recent years has brought about a surprising renaissance for non-motorized traffic in China.A large number of non-motorized vehicles have to mix in non-motorized lanes,leading to a more complex operational condition of urban traffic.The current infrastructures and administrative measures for non-motorized vehicles could not satisfy the needs of the rapid development of non-motorized traffic,which might cause potential threats to safety and operational efficiency of urban traffic in China.Therefore,based on the formation mechanism of road behaviors of non-motorized vehicle users and the variety of behavioral modes among the group,a microscopic simulation model for mixed non-motorized flow was developed to get a better understanding of the characteristics of nonmotorized flow at road segments and unsignalized intersections in cities.The main research contents and conclusions are summarized as follows.First,a video stabilization algorithm was developed to improve the quality of field video data of non-motorized flow,and a method combining the Gaussian Mixture Model and the Kalman Filter was employed to extract trajectories of non-motorized vehicles from video recordings.Afterwards,several kinematic parameters were identified as features characterizing behavioral modes from the reduced dimensional real-world trajectory data.The heterogeneity of road behaviors among nonmotorized vehicle users was confirmed through the clustering analysis of those kinematic parameters,and the non-motorized vehicle users were then characterized into three groups according to their behavioral style on the road,including careful riders,average riders and aberrant riders.Second,to better depict the traffic environment that non-motorized vehicle users were exposed to,the environmental clues in the traffic environment were classified into static obstacle information,perceived difficulty,perceived risk and acoustic information(belling)according to the perceptual characteristics of road users.After the key influence factors of each category of environmental information were discussed and summarized,an environmental perception model with a basic motion model was constructed to simulate the information collection process of non-motorized vehicle users in the traffic environment,and the algorithm implementation of the model in MATLAB was also presented.Third,fuzzy logic methods were used to simulate the information-processing procedure of nonmotorized vehicle users,and a fuzzy-rule based decision model was proposed to model the decisionmaking process of instantaneous route choice and belling intention for different types of nonmotorized vehicle users,which were considered as the embodiment of the behavioral heterogeneity among them.A “Riding Style Questionnaire of Non-motorized Vehicle Users”(RSDNV)was developed as a quantitatively measurement tool of non-motorized vehicle users' behavioral preference,and a questionnaire score-based classification method of one's riding style was constructed and validated through comparison with the behavioral style identified based on field data.Then,a questionnaire survey including the RSDNV and questions related to the fuzzy rules to be defined was conducted among a sample of non-motorized vehicle users.The behavioral styles of the respondents were determined based on the RSDNV scores,and the fuzzy rules were then constructed according to the answers of each category of respondents.Fourth,according to the interaction mechanism between non-motorized vehicle users and the traffic environment,a modified social force model for non-motorized vehicles was constructed when taking road users' perceptual and decision-making properties into consideration and then programmed in MATLAB.Based on the simulation procedure,a series of simulation experiments were performed under the condition of different simulation parameter combinations to explore the characteristics of non-motorized flow at road segments and unsignalized intersections as well as the formation of some microscopic phenomenon in the traffic flow.Theoretically,the current results add to the existing system of microscopic simulation of nonmotorized flow.Practically,this study not only provides a theoretical basis for the design of evidencebased administration and development measures for urban non-motorized traffic,but also could facilitate the promotion of traffic safety and the optimization of traffic environment for non-motorized vehicle users in China.