Zhao Ziwen, Feng Tianjun, Qu Jinghuan
Jilin Jianzhu University
Abstract: In order to analyze the gaze characteristics of urban road drivers in cold area congestion, the German Dikablis eye tracker and its supporting D-Lab software were used to carry out real vehicle tests on urban roads in Changchun City, and the influence of cold area congestion road characteristics on driver gaze characteristics was quantified by statistical analysis. The results show that drivers mainly obtain traffic information through gaze when driving, and the fixation points are mostly concentrated in road conditions and road traffic flow, especially in the state of ice and snow thawing pavement. The one-way variance result of driver fixation duration was less than 0.05, which was significantly different when driving on urban roads in cold areas under congestion.
Keywords: gaze characteristics; congestion state; cold zone; urban roads;
About author:Zhao Ziwen (1997—), female, from Linyi, Shandong, master's student, research direction is road traffic safety.
Fund: Science and Technology Research Program of Jilin Provincial Department of Education, Project No.: JJKH20210270KJ; Jilin Province College Students Innovation and Entrepreneurship Training Program, Project No.: 202010191037;
introduction
With the improvement of social and economic development and living standards, the number of motor vehicles in the mainland has increased year by year, according to data from the Ministry of Public Security, the number of motor vehicles in China in 2020 was 372 million, of which 281 million were automobiles, an increase of 7.5% year-on-year. While motor vehicles provide people with convenience, they also create a series of problems, such as causing congestion and causing traffic accidents.
The cold region of the northern part of the mainland has a long winter, frequent snowfall, frequent snow and ice, and the impact of winter weather conditions on road traffic is prominent, which is easy to cause traffic congestion, resulting in changes in driver emotions and traffic accidents. Emotional changes are mainly reflected in visual changes, and gaze behavior is the most important way for drivers to obtain road and traffic information [1], and the driver's gaze behavior has attracted the attention of scholars at home and abroad.
YEKHSHAYAN AND LEE [2] USED SIMULATED DRIVERS TO ANALYZE THE CORRELATION OF EYE TRACKING GAZE BEHAVIOR AND STUDIED THE DEGREE AND TYPE OF DISTRACTION. TATLER et al. [3] studied the visual transfer mode of drivers under free perspective, studied the driver's gaze allocation, and obtained the mechanism and emerging theoretical framework of controlling visual distribution. Hu Hongrui et al. [4] simulated different environments through virtual driving software, believing that under the same road conditions, the worse the light, the more the driver's gaze and echo points. Hu Yueqi et al. [5] studied the influence of driving experience and environment on driving safety, and the results showed that there were obvious differences in the visual characteristics of drivers in ordinary road sections and extra-long tunnel sections. Ye Jianguang [6] concluded that in the environment of mountain expressways, the driver's fixation duration is shorter and the cognitive load is smaller, which is generally manifested as more long-term fixation in the obscured space than in the open space.
At present, the road driving environment studied mainly focuses on tunnels, mountain roads and highways, but there is less research on urban roads in complex environments. As an important part of the transportation system, the impact of urban roads on drivers has gradually attracted the attention of researchers [7].
1 Characteristics of pavement conditions in cold areas
The main feature of cities in cold regions is that the ice and snow period is long, and under frequent snowfall weather conditions, urban roads will produce snow and cause road surface to freeze, resulting in a decrease in the road adhesion coefficient. With the passage of snowfall time and the change of temperature and light, the ice and snow status of the road surface will also change, and the ice and snow status of the road surface in winter in the cold area can be divided into: (1) natural snow. The snow covered on the road after snowfall has a small density, about 0.25~0.5 g/cm3, which generally occurs during snowfall and a short time after snowfall. (2) Smooth and compacted snow. The natural snow on the road is crushed by vehicles and pedestrians, and the surface begins to become smooth, and the density increases compared with the natural snow state, between 0.5~0.6 g/cm3. (3) Thaw ice and snow. Affected by the continuous low temperature, the road snow is melted by sunlight and vehicle tire friction, and condenses into ice film in combination with the road surface, and the ice film density reaches 1.0 g/cm3, the road surface becomes very smooth, and the driving vehicles are prone to side-skid and rear-end collision accidents. (4) Melt ice and snow. When the temperature rises to 0°C and above, the road ice and snow begin to melt, and the state changes to melting ice and snow. Affected by light and urban road traffic, molten ice and snow generally appear on both sides of the road, and are in a discontinuous dry and wet state. Different road surface ice and snow conditions correspond to different road adhesion coefficients [8], see Table 1.
Table 1 Range of pavement adhesion coefficient Download the original figure
It can be seen that the adhesion coefficient of the pavement is very different under ice and snow conditions and under normal conditions.
2 Experimental protocol design
2.1 Test section and time determination
Changchun urban roads were selected as the experimental sections, including expressways, trunk roads and other urban road types, with a total distance of 11.5 km. The test route passes through commercial areas, schools, residential areas, etc., and the traffic flow changes significantly, and is subject to lateral interference by pedestrians and vehicles, which has typical urban road characteristics. The specific route is shown in Figure 1, starting from Renmin Street auxiliary road, turn right at the intersection through Freedom Road to Yatai Avenue, then turn to Jiefang Road, enter Xinmin Avenue along Jiefang Road, turn into Guilin Road at the intersection of Xinmin Avenue and Guilin Road, and then enter Freedom Avenue through Peony Street, continue to drive to Renmin Street to reach the end.
Figure 1 Test route Download the original image
In 2020, the peak travel delay index of Changchun's road network was 1.836. In order to study the fixation characteristics under congestion, the peak time of traffic flow during the ice and snow period was selected as 17 00-18 30, and the test was carried out in the state of ice-free and snow-free, smooth compacted snow, thawed ice and snow, and ice and snow on the test route, and the pavement friction coefficients were 0.8~0.9, 0.2~0.25, 0.10~0.20, and 0.30~0.40, respectively.
2.2 Test instrument and personnel selection
The test used Dikablis eye tracker, D-lab software, forward selection inverter, camera, small car and friction coefficient meter, some of which are shown in Figure 2 and Figure 3. The eye tracker was developed by German Ergoneers and has the characteristics of high precision and low invasiveness. Dikablis is equipped with two cameras: eye-camera and field-camera, the former is responsible for tracking the eyes, the latter is responsible for recording the scene, which can collect eye movement data and real scene data at the same time, and can analyze eye movement data from all angles and angles with D-lab software.
Figure 2 Test instrument Download the original figure
Figure 3 Test supporting software Download the original image
Select 5 skilled drivers with more than 5 A driving experience, including 3 men and 2 women, who have normal visual function, no physical defects and experience of serious accidents. In order to ensure the normal conduct of the test and the safety of personnel, the driver is required to be in a good mental state before the test, ensure adequate sleep and no drunk driving behavior.
2.3 Test process conducted
(1) The test personnel conduct field inspections of the entire selected urban roads, and train the recruited drivers in the use of instruments to ensure that the test is carried out normally. (2) Before the start of each test, the tester shall measure and record the friction coefficient of the pavement of the test section. (3) The driver wears the eye tracker, the observer starts the computer and calibrates the eye tracker, and after calibration, the driver starts the test vehicle for testing. (4) During driving, the observer uses the D-Lab software to observe the driver's viewpoint capture to ensure that the instrument data is valid. (5) After each driver drives the test section, the observer exports and stores his gaze characteristic data and video data. The experiment was repeated, and a total of 20 sets of test data were obtained; (6) After the test is completed, the tester integrates the recorded data, and classifies and preprocesses the eye tracking data according to the pavement condition.
3 Gaze characterization
3.1 Gaze time
Gaze time refers to the period of time between the two jumping movements of the eyeball, and the proportion of fixation time obtained by statistically measuring the test eye movement behavior data is shown in Figure 4.
Figure 4 Percentage of driver gaze time/% Download the original figure
It can be seen from Figure 4: (1) No matter what kind of road surface state, when drivers drive on urban congested roads, they mainly obtain road information through gaze behavior, and their fixation time accounts for more than 70%. (2) With the decrease of the road adhesion coefficient, the proportion of average fixation time of drivers shows an increasing trend, which affects their ability to obtain road information.
3.2 Foveated distribution
The distribution of fixation points can be divided according to the fixation area, but this method can only obtain the driver's concentration of gaze on a certain area, and cannot reflect the gaze interest in specific things [9]. The distribution of driver gaze points is divided by gaze object properties. By observing the video recorded by the eye tracker in the test, the classification of the driver's main gaze points when driving is summarized.
Table 2 Foveated classification Download the original image
According to the classification statistics, the cumulative number of fixations of drivers in the state of no ice and snow, smooth compacted snow, thawed ice and snow, and melted ice and snow on congested roads is shown in Figure 5.
Figure 5 Proportion of cumulative fixation times for various fixation points Download the original figure
It can be seen from Figure 5 that when drivers drive on urban roads, their fixation points are mainly concentrated in road traffic flow and road conditions, especially when thawing ice and snow, drivers have a higher number of gazes on these two types, accounting for more than 40%. In ice and snow conditions, drivers pay less attention to the road landscape, which shows that drivers maintain a high degree of tension when driving in congested ice and snow conditions, and have no time to pay attention to the road landscape. Under the three ice and snow conditions, the number of times drivers look at the dashboard and rearview mirrors in the car is relatively average, accounting for about 7%.
3.3 Gaze duration
Gaze duration refers to the duration of the time that the center position of the visual axis remains unchanged during fixation, measured in ms. The length of gaze duration indicates how easy the driver needs to process information, and longer fixation duration indicates more difficult information to process, but sometimes a target with too long fixation duration is not necessarily more meaningful than a target with a short fixation duration. The statistical analysis of the data of drivers congesting urban roads in cold regions is shown in Table 3.
Table 3 Gaze duration statistics Download the original figure
As can be seen from Table 3, the average fixation duration of drivers driving on congested roads in thawed ice and snow is the highest, reaching 327.664 ms. The adhesion coefficient in the thawed ice and snow state is extremely low, and the adhesion cannot be balanced with the external force, resulting in the destruction of vehicle stability and prone to side skidding. This is where drivers are the most complex and difficult to process road information. In addition, because molten ice and snow are common on both sides of the road and show the characteristics of dry and wet, this state has little impact on drivers in urban congested roads, and the average fixation duration is very different from that of the road surface without ice and snow, and the distribution is about 300 ms.
In order to further judge whether the fixation duration is different between the road states, the samples are first tested for normality, and the S-W statistic and the corresponding probability P value of the test result are shown in Table 4.
Table 4 Gaze duration S-W test statistics Download the original figure
From the test results, it can be obtained that the average fixation duration P value of drivers under different road surface conditions in cold areas is greater than 0.05, which is in line with the normal distribution. The homogeneity of variance test was performed on the four groups of sample data, and the Levene statistic based on the mean value was obtained as 1.391, corresponding to the significance P=0.920>0.05, which can be considered that the number of samples in different groups of pavement conditions in the cold region has the homogeneity of variance.
Further one-way ANOVA analysis shows that the significance P=0.001<0.05 is obtained, that is, the average fixation duration of drivers when driving on urban roads in cold areas under congestion is significantly different. The average fixation duration under different ice and snow pavement conditions was compared by the LSD method, and the multiple comparison results obtained are shown in Table 5.
Table 5 Gaze duration LDS test results Download the original figure
It can be seen from Table 5 that there is no significant difference between the pavement without ice and snow and the dissolved ice and snow, and there is a significant difference between the other two pavement state samples.
4 Conclusion
Starting from the driver's gaze behavior, the influence of urban road congestion in cold area on the driver's gaze characteristics was analyzed. Through descriptive statistical analysis, the distribution law of driver's fixation behavior under different road surface conditions was obtained, and the proportion of fixation time increased with the decrease of road adhesion coefficient, up to 76.65%. In congested roads, smooth compaction snow and thaw, drivers pay more attention to road conditions and traffic flow. The normal test and differential analysis of the fixation characteristics were carried out to determine that the duration of the driver's fixation was affected by the congestion environment in the cold area, and the road adhesion coefficient < 0.25, which caused the driver to obtain road information and affected driving safety. According to this result, the driver's gaze characteristics can be used as an evaluation index to evaluate the driving safety of congested urban roads in cold areas. According to the climate characteristics of cold areas, ensure timely and effective winter snow removal work, and reduce the adverse effects of ice and snow weather and congestion on urban roads in cold areas.
bibliography
[1] LI Zuomin. Traffic Engineering[M].2nd ed. Beijing:China Communications Press,2000.
[2] YEKHSHATYAN L, LEE J D. Changes in the correlation between eye and steering movements indicate driver distraction[J]. IEEE Transactions on Intelligent Transportation Systems, 2013, 14(1):136-145.
[3] TATLER B W, HAYHOE, MARY M, et al. Eye guidance in natural vision:reinterpreting salience[J]. Journal of vision,2011, 11(5):5.
HU Hongrui, LIU Longfan, XIONG Yan, et al. Biomedical Engineering Research, 2016, 35(4):255-259.)
[5] HU Yueqi, LIU Haoxue, ZHU Tong, et al. Research on driver's visual characteristics in expressway special tunnel environment[J].China Safety Science Journal, 2017, 27(6):31-36.)
[6] Ye Jianguang. Research on eye movement and psychological characteristics of highway drivers in mountainous areas and its application in landscape construction[D].Chongqing:Chongqing Jiaotong University, 2017.)
[7] ZHU Dianchen. Study on the influence of expressway ramp environment on the physiological characteristics and behavior of drivers[D].Guangzhou:South China University of Technology, 2018.)
[8] Shi Yalei. Research on highway traffic control under ice and snow conditions[D].Tianjin:Hebei University of Technology, 2018.)
[9] TAN Ting, WANG Yuchen, ZONG Chenhong, et al. Study on the difference of visual characteristics of drivers in different sections of highway tunnel group[J].Logistics Science and Technology, 2020, 43(10):68-72, 82.)
Disclaimer: We respect originality and focus on sharing. Some of the content comes from the Internet, the copyright belongs to the original author, only for learning reference, prohibited for commercial use, such as inadvertently infringing the intellectual property rights of which media, company, enterprise or individual, etc., please contact to delete (email: [email protected]), and this headline push content only represents the author's views, and the headline number operator has nothing to do, the authenticity of the content please readers to identify by themselves, this headline number does not bear any responsibility.