Study On Visual Behavior Of Non-motorized Cyclists Under Complex Road Conditions

The driving behavior of non-motorized cyclists is one of the most important factors affecting road safety.Vision is the most important way for non-motorized cyclists to perceive traffic information and perceive potential dangers.Since the dynamic traffic environment contains much traffic information that can greatly affect the visual search behavior and riding behavior of cyclists,it is necessary to analyze the visual behavior of non-motorized cyclists under complex road conditions can improve non-motorized traffic safety.The following work is done to study the visual behavior of cyclists in different traffic conditions.First,this study clarifies the complex conditions of non-motorized traffic and the travel characteristics of cyclists,then introduces the visual acquisition technology methods and the basic forms and characterization parameters of the visual behaviors of non-motorized cyclists,and also clarifies a number of eye-movement characteristics data such as gaze behavior,sweeping behavior,blinking behavior and pupil diameter change as the basic indicators of cyclists’ visual behaviors.In this paper,a typical city road in Nanjing contains city arterial roads,cross intersections and Tintersections as the experimental scenario.Twenty subjects are chosen to conduct real vehicle experiments using Tobii Glasses2 eye-tracking device,and the SPSS software and Ergo LAB humancomputer interaction system are used to process the experimental data.Second,to find out the eyemovement characteristics and visual load change patterns of cyclists under complex traffic conditions by comparing and analyzing the variability of several eye-movement data such as cyclists’ gaze time,gaze number,gaze area and target under different traffic flow and different bicycle type conditions.Third,for the cyclists’ gaze shifting behavior pattern,Markov chain theory method is used to divide the cyclists into two groups according to the non-motorized traffic and count the distribution of cyclists’ gaze points respectively,then the k-means clustering method is used to divide the cyclists’ gaze interest areas,and finally the one-step shifting gaze probability distribution and the gaze probability smooth distribution of the two groups of cyclists are calculated.In addition,in order to investigate the internal drivers of the gaze shifting behavior patterns of non-motorized cyclists,the gaze shifting correlation matrix is built by imitating the gray correlation model to analyze the influence of other factors on the gaze shifting probability results.Finally,based on the differences of cyclists’ visual behaviors under different traffic conditions,eight eye-movement indicators are selected to build the original matrix,and the visual concentration of cyclists is evaluated by principal component analysis.Then,the concentration level is classified into three categories: "concentration type,divergent type,and distracted type" based on the overall score of the evaluation,and the factors influencing the cyclists’ concentration level are analyzed by using an ordered Logistic regression model.The results show that(1)non-motorized cyclists pay most attention to motor vehicle information at signalized intersections,and in the horizontal direction of the gaze area,e-bike riders focus on the area in front of the current lane,while rickshaw riders mainly focus on the middle area,taking into account the right area.In the vertical direction,the number of bicyclists’ gaze points on the signal area was higher.(2)To compare the blink and pupil data of different types of bicyclists,it is found that e-bike riders are more susceptible to visual pressure,while rickshaw riders have higher visual fatigue;(3)To compare the gaze shifting behavior patterns of riders under two traffic flow conditions,in which the peak group’s gaze shifting behavior is concentrated and stable,while the flat group’s gaze shifting pattern is more flexible and can take into account the left and right(4)Nonmotorized traffic flow and crossing waiting time are significant factors affecting cyclists’ concentration,and bicycle type also affects cyclists’ concentration,and the accuracy of classification evaluation is 65%,logistic regression classification is feasible.Based on these results,the eye-movement characteristics,gaze shifting behavior patterns and factors influencing the degree of cycling concentration of non-motorized cyclists under complex road conditions are derived.The results can provide theoretical support for non-motorized traffic safety research.