Eye Movement Characteristics Of Non-Motorized Cyclists In Low Visibility Conditions

Eye movement characteristics are one of the research directions of the driving behavior of road traffic participants,non-motorized vehicles are an important part of road traffic participants,visibility is an important factor that interferes with driving behavior,especially in low visibility environments,the eye movement characteristics of non-motorized cyclists can be greatly affected,leading to traffic accidents and causing road traffic safety problems,so it is necessary to study the oculomotor characteristics of non-motorized cyclists in low visibility.Using oculomotor characteristics as a starting point,analyze the oculomotor characteristics of cyclists in different visibility conditions.Firstly,representative signalized intersections and physically segregated non-motorized lanes on urban roads were selected as two experimental sites,and Tobii Glasses eye trackers were used to conduct real-world experiments.Eye tracking data were collected from 25 non-motorized cyclists under high visibility,lower visibility and low visibility conditions,and data were exported and pre-processed.Secondly,the cyclists’ gaze behavior,sweeping behavior,blink movements and pupil diameter changes were statistically compared at different visibility levels to analyze the cyclists’ oculomotor characteristics and the degree of visual load.Thirdly,delineate the cyclists’ gaze interest areas,Markov theory was applied to obtain the cyclists’ gaze shift characteristics,and build a visual change rate model to analyze the cyclists’ visual decline when visibility was reduced.Fourthly,principal component analysis was chosen to obtain the five important eye-movement indicators that influence cyclists’ visual behaviour.,fuzzy clustering was applied to evaluate cyclists’ visual search patterns in low visibility,individual differences were considered to determine the visual search pattern classification,one-way ANOVA was used to test the significance of the classification results.The results showed that:(1)There were differences in cyclists’ eye movement data in different visibility levels.As visibility decreased,cyclists’ gaze point distribution was more dispersed,gaze duration increased and visual load level increased.(2)Cyclists’ gaze areas at intersections were divided into areas above,ahead,to the left and to the right,while in straight ahead conditions the gaze areas were divided into distant areas,areas ahead,motor vehicle areas,road and building areas and pedestrian areas.(3)The highest probability of gaze shift at intersections is to stay in the front area,and as visibility decreases,the probability of gaze shift in the left area decreases and increases in the right area;the highest probability of gaze shift on straight roads is to stay in the far area,followed by a shift to the front area,and the probability of gaze shift in the motor vehicle area,road and building area and pedestrian area gradually increases.The probability of shifting gradually increases in the area of motor vehicles,road buildings and pedestrians.(4)The five indicators of blink,sweep,mean sweep time,pupil diameter variance and mean blink time were selected by principal component analysis to explain 97.686% of the original data.Using fuzzy theory to determine the best classification threshold,the visual search patterns of the selected experimental samples were classified into five categories: "excellent","good","moderate","qualified" and "unqualified",The F-statistic was used to test the significance of the five categories,and it was found that the five indicators of each category were all less than 0.05,which was a significant difference,and the classification results were determined to be reasonable.By studying the eye-movement data of non-motorized cyclists under different visibility levels and analyzing the visual change patterns under different visibility,we can provide theoretical support for safe cycling in low visibility and for the design of urban road facilities.