Impact of Urban Undersea Tunnel Longitudinal Slope on the Visual Characteristics of Drivers

-is study aims to investigate the impact of the urban undersea tunnel longitudinal slope on the visual characteristics of drivers. 20 drivers were enrolled to conduct the real vehicle test of the urban undersea tunnel. First, the data of average fixation time and visual lobe were collected by an eye tracker. -e differential significance was tested using the one-way repeated measures analysis of variance (ANOVA).-en, the difference between the up-and-down slope (direction) factor and the longitudinal slope (percent) factor on the two indexes were analyzed using the two-way repeated measures ANOVA. Second, by constructing a Lorentz model, the impact of the longitudinal slope on the average fixation time and the visual lobe were analyzed. Besides, a three-dimensional model of the longitudinal slope, average fixation time, and visual lobe was quantified. -e results showed that the average fixation time and visual lobe under different longitudinal slopes markedly differed when driving on the uphill and downhill sections. -e average fixation time and visual lobe under two factors were markedly different. Moreover, with an increase in the longitudinal slope, the average fixation time exhibited a trend of increasing first then decreasing; the visual lobe exhibited a trend of decreasing first and then increasing. -e average fixation time reached the minimum and maximum value when the slope was 2.15% and 4.0%, whereas the visual lobe reached the maximum and minimum value when the slope was 2.88% and 4.0%. Overall, the longitudinal slope exerted a great impact on the visual load of the driver.


Introduction
Lately, the number of urban tunnels crossing rivers and mountains has gradually increased to adapt to the rapid development of the urban economy and regional expansion [1]. In coastal cities, the restrictions on the sea area have deferred the development of cities. us, in recent years, numerous special long undersea tunnels have been built, such as the Xiamen Xiang'an Tunnel (full length: 8.69 km) and Qingdao Jiaozhou Bay Tunnel (full length: 7.80 km), to accelerate regional economic development and shorten mileage. Typically, undersea tunnels have the characteristics of long distance, large longitudinal slope (usually 3%-4%), and high brightness of the hole (the effect of "black and white hole" is notable) [2]. us, compared with the general roads, undersea tunnels affect drivers' physical and psychological burdens in terms of long large uphill and downhill slopes and poor sight [3]. us, exploring the visual characteristics of drivers in long large uphill and downhill slopes of undersea tunnels is of great significance to improve the traffic safety.
Currently, relevant research has not been conducted on the visual characteristics of drivers in the long and large slopes of undersea tunnel, and most studies have focused on highway tunnels and under-crossing tunnels. Regarding highway tunnels, Pan et al. analyzed the visual obstacles and driving visual comfort of drivers in highway tunnel entrance and exit using the pupil area maximum instantaneous velocity value (MTPA) as a visually compliant evaluation index [4]. Redenbo et al. and Macdonald et al. realized that the visual load was determined by both the visual physiological load and the visual perception load [5,6]. Green et al. analyzed the factors influencing tunnel sight distance based on visual principles [7]. e study by Manser and Hancock noted that drivers' perception of speed and their subsequent behavior will be affected by the visual mode due to the influence of the tunnel sidewall effect [8]. Regarding urban tunnels, based on real vehicle tests, Feng et al. proposed that drivers displayed the highest growth rate of the heart rate when driving uphill (longitudinal slope range, 3.5%-4.0%) and downhill (longitudinal slope range, 4.0%-4.5%) [9]. By comparing the visual information load of drivers on the ground and underground with the characteristics of vehicle operation, Chen and Zheng analyzed the influence mechanism of the visual environment of the underground road on the visual information load of the drivers [10]. By developing a multiple linear regression model among the heart rate, illumination, and vehicle speed at the entrance segment, Wu et al. investigated the impact of external factors on the heart rate [11]. Regarding long and large downhills, Yang et al. revealed that the drivers' pupil area was larger than other road sections in the black spot section of the long downhill road; the pupil area could effectively characterize the drivers' psychological stress, and the drivers were more sensitive to the left turn [12].
rough three different models, Gu revealed that when the downhill longitudinal slope was 3.7%, the eye movement speed was the smallest, and when the downhill longitudinal slope was 3.6%, the range of the gaze point was the smallest [13].
Above existing studies primarily focus on the analysis of the characteristics of the drivers' eye movement such as the gaze, saccade, and pupil changes in highway tunnels [14,15] and urban tunnels [16,17]. Unfortunately, relevant studies on the drivers' visual characteristics in urban undersea tunnels is lacking at this stage.
us, this study aims to analyze the drivers' visual characteristics at long and large downhill sections in urban undersea tunnel and provide an effective theoretical basis for the safe operation of urban undersea tunnels.
In this study, the drivers' eye visual data were collected in the Jiaozhou Bay undersea tunnel via the real vehicle test, and the drivers average fixation time and visual lobe data were analyzed. Besides, the correlation among longitudinal slope, average fixation time, and visual lobe was explored to assess the visual load of the drivers in the undersea tunnel section. To achieve the objectives of this study, the following issues must be addressed: (1) Does a significant difference exist in the driver visual data when driving on a different longitudinal slope in the tunnel? (2) Does a quantitative relationship exist between the longitudinal slope and the visual parameters such as the average fixation time and the visual lobe? (3) Is there any difference in the eye movement behavior and psychological load when driving at the uphill and the downhill section?

Test Road.
In the real vehicle test described in this study, the Qingdao Jiaozhou Bay Tunnel was selected as the research object (Table 1). It is a divided tunnel, and cars in different directions do not interfere with each other. e starting and ending points of the test were the Huangdao Toll Station of the Jiaozhou Bay Tunnel.

2.2.
Participants. e participants were required to have, at least, 3-year experience in the driving age, owing to the particularity of urban undersea tunnels and to ensure the safety. Besides, the participants were not achromatopsia and hypochromatopsia, and their visual acuity was normal or corrected-to-normal; the participants had no psychological and physical defects; and the participants did not experience any major traffic safety accidents. Overall, 20 participants (14 males and 6 females) were included in the study. eir average age was 33.850 years (SD � 6.952), while their average driving experience was 6.905 years (SD � 3.987). All participants got paid (300 RMB) after the tests. Table 2 presents the basic information about the participants.

Test Equipment and Vehicle.
e eye tracker used in the tests was a head-mounted Dikablis Professional eye tracker (Ergoneers GmbH, Geretsried, Germany) with the following characteristics: sampling rate, 60 Hz; pupil tracking accuracy, 0.05; sight tracking accuracy, 0.1°-0.3°; scene camera, 1920 × 1080 Full HD; and scene camera range, 40°-90°. e eye data collection and postprocessing of the participants were performed by the equipped D-Lab software version 3.54 (Ergoneers GmbH, Geretsried, Germany). Eye movement recording was performed on a Dell laptop with Intel Core i7 2.7 GHz CPU, 16 GB memory, Nvidia Quadro M200M graphics card 1920 × 1080 pixels of screen resolution, and Windows 10 operating system. e test vehicle was a Chevrolet Cruze automatic 5-seater; and other devices included a Digital Video and car power supplies. Figure 1 shows the eye tracker and vehicle.

Test Method.
To avoid the impact of subjective deviation and personal preference in the consciousness of experimenters and participants on the trial results, we adopted the double-blind principle and tried to evade other factors affecting the participants driving. Before the test, the drivers were informed with passing the test section based on their usual driving habits and following the traffic rules to freely travel the uphill and downhill tunnel sections. e specific steps of the test were as follows: (1) e experimenter drives the test vehicle to the starting point and installs the equipment such as power supply and driving recorder. (2) Before the formal test, the experimenter selects one participant to perform the pretest according to the established test plan to ensure that the test process is correct before conducting the formal test. If the equipment failed to work normally, the test data collection was stopped; after returning to the starting point, the participant was allowed to close eyes for rest in 5 mins before restart the test. (6) To guarantee the reliability of test data, each participant repeated four times; every participant drove in the middle lane with the free-flow speed of 70-80 km/h; the tests were performed during lowpeak periods at night (21:00-23:00).

Index
Selection. e average fixation time and the visual lobe were selected to assess the driver visual characteristics in the long and large slope environment of an undersea tunnel.
e visual parameters were obtained in different longitudinal slope environments to reflect the information extraction, collection ability, and visual load situation of the drivers [18].
(1) Average fixation time: e average fixation time is the average value of the driver's total fixation time divided by the number of fixation times in a certain period, also being a crucial index of visual characteristics. e long average fixation time suggests the more difficult the visual information, the more the driver's eye needs to collect [19].
(2) Visual lobe: e magnitude of the jump from the viewpoint between two adjacent gazes is defined as the saccade distance, being usually presented in degrees of the angle of view. In psychology, the concept of the visual lobe denotes the size of the visual recognition scope. It is easier to visualize when the search target exists within the scope of the visual lobe centered on the previous gaze point, while the target beyond this scope is difficult to visualize. Furthermore, the median value of the saccade distance is used to measure the visual lobe [20].

Data Collection and Preprocessing.
We contacted the construction unit in advance to obtain the alignments. Before the test, the alignment data were combined with driving video to find some points of slope change in advance. After all the tests were completed, the task segments were divided in the D-Lab software. According to the information of mileages, slopes, markings, and signs, the points of slope change were identified to divide the slope task segments.
In order to prevent invalid data from affecting the results, the collected data were processed by the Pauta criterion. e calculation method of Pauta criterion was presented as follows: where x i denotes the test data, x denotes the mean value of the data samples, and σ denotes the standard deviation of the data samples. e threshold of the eye tracker for determining the fixation is 120 ms.

Significant Difference Analysis.
e task segments under different longitudinal slopes of the uphill and downhill sections were divided in the eye tracker supporting software D-lab. Finally, the visual data under different longitudinal slopes of the uphill and downhill sections were derived by the software. Table 3 shows the mean and standard deviation of the average fixation time and the visual lobe under different longitudinal slopes of the uphill and downhill sections. Tables 4 and 5 show the results of one-way repeated measures ANOVA. Figure 2 presents a box line diagram of the average fixation time and visual lobe under different longitudinal slopes of the uphill and downhill sections. All data satisfied the Kolmogorov-Smirnov normality test (P > 0.05).
(a) Average fixation time: First, based on the one-way repeated measures ANOVA, a significant difference test was performed on the average fixation time. It can be seen from the column Prob > ChiSq (
e assumption of sphericity has been satisfied. It can be seen, from the column sphericity assumed ( To further explore the impact on the average fixation time and visual lobe under the two factors of up-and-down slope and longitudinal slope, the significant difference levels of the two factors and the significant difference level between them were compared. All data satisfied the normality test by the K-S normality test (P > 0.05).From Tables 6 and 7 Combined with the overall mean values presented in Table 3, the average fixation time under the uphill factor was significantly smaller than the average fixation time under the downhill factor; the visual lobe under the uphill factor was significantly higher than the downhill factor.

Model Fitting Analysis.
To elucidate the correlation between the average fixation time and the visual lobe under different longitudinal slopes in the uphill and downhill sections, the data were analyzed by the curve fitting analysis via the OriginLab data analysis software. Fitting and comparison revealed that the Lorentz model has a good fitting effect, and the fitting formula is as follows: where y is the dependent variable, suggesting the average fixation time or visual acuity; x is an independent variable, denoting the slope of the reaction; y 0 , A, π, w, and x c are constants; and x c is the positional parameter, denoting the position of the average fixation time or peak of the visual lobe under the longitudinal slope. Figure 3 shows the derived fitted curve of the average fixation time and the visual lobe. Based on Table 8, the judgment coefficient R 2 and the corresponding P-value indicated that the fitting degree of the curves were excellent. Figure 3 and Table 8  (2) When a driver drove in this tunnel, the visual lobe usually decreased first but then increased with the longitudinal slope. When the longitudinal slope was about 2.685% in the downhill section, the visual lobe reached a maximum of 5.018°. Furthermore, when the longitudinal slope was about 2.916% in the uphill section, the visual lobe reached a maximum of 3.456°.
To illustrate the correlation among the longitudinal slope, the average fixation time, and the visual lobe, the data were analyzed using a 3D surface fitting analysis by using the OriginLab data analysis software. Fitting and comparison revealed that the Gauss2D model exhibited a good fitting effect. e fitting formula is as follows:  Table 9, the judgment coefficient R 2 of the fitting curve and the corresponding P value indicated that the curve exhibited a good fitting degree. Figure 4 and Table 6 revealed the following:

Discussion
(1) e experimental data analysis focused on the relationship between the two indexes of average fixation time and visual lobe with the longitudinal slope. It is generally believed that, as the slope increases, the driver's visual performance is worse or the visual load is greater. However, Feng et al. [16] has discovered that the relationship between slope and heart rate growth was not linear. Heart rate growth reached the maximum at 3.33% and 2.91% of the uphill and downhill, respectively. However, a peak existed in the results of this test; thus, the longitudinal slope could be divided into three intervals by the regularity changing: slope <2%, 2% ≤ slope ≤3% (peak area), and slope >3%. rough the speed observation, the overall speed difference of the three sections was not significantly different (uphill, p � 0.055 > 0.05; downhill, p � 0.078 > 0.05). (1) In the interval where the longitudinal slope value was >3%, the average fixation time was the largest, and visual lobe was smaller. is section (especially the longitudinal slope was about 4%) was generally the tunnel entrance or nearly the entrance. Typically, vehicle speed increases in the long downhill section of tunnel entrance, and the drivers may need to step on the brakes [21] and stare at the dashboard to avoid speeding and rear-end collision. Meanwhile, for the uphill section, the drivers need to step on the throttle and gaze at the dashboard and rearview mirror because of the low speed in uphill section. us, the driver had numerous driving tasks and too much visual load on the large longitudinal slope section. In addition, the illuminance and the cross-section of the space that were changing dramatically at the entrance of the tunnel were highly dangerous or be psychologically nervous for drivers, which ultimately led to a marked reduction in driving comfort [15]. (2) In the interval where the longitudinal slope value was <2%, the driver's average fixation time was relatively large and visual lobe was smaller; this section was generally located in the middle of the tunnel, wherein the driving task and visual load were the smallest. However, the driver might experience driving fatigue when driving on a gentle section for a long time [22].
(3) In the interval where the longitudinal slope value was between 2% and 3%, the driver's average fixation time was the smallest and visual lobe was the largest; this section was generally located after a slope of >3% and <2%; the driver would be relieved from high visual load or driving fatigue. Furthermore, the slope perceived by the driver was just adapted to the driving state, at which time the driver was in a relatively low driving load state.  (2) In the highway tunnels, due to changes in illumination and space environment, accidents are generally concentrated at the entrance and exit of the tunnel [23,24]. e entrance of undersea tunnel may be more dangerous due to the larger longitudinal slope.
However, the dangerous section inside the undersea tunnel is ignored. For example, the fatigue of driving in a monotonous tunnel interior environment for a long time [25] may lead to traffic accidents. Although there have been many safety improvement methods in

Conclusions
(1) When a driver drove in the uphill section of the urban undersea tunnel, the average fixation time and the visual lobe were different under different longitudinal slope conditions. e average fixation time in the uphill section tended to decrease first but then increase with the longitudinal slope, whereas the visual lobe increased first and then decreased with the longitudinal slope. In the uphill section of the tunnel, the visual load was moderate at 2.47% and 2.88% of the longitudinal slope, and the driver's  visual load was the highest at 4.0% of the longitudinal slope.
(2) When a driver drove in the downhill section of the urban undersea tunnel, the average fixation time and the visual lobe were different under different longitudinal slope conditions. e average fixation time in the downhill section tended to decrease first but then increased with the longitudinal slope, whereas the visual lobe increased first and then decreased with the longitudinal slope. In the downhill section of the tunnel, the visual load was moderate at 2.15% and 2.88% of the longitudinal slope, and the driver's visual load was the highest at 4.0% of the longitudinal slope.
(3) When a driver drove in the urban undersea tunnel, the average fixation time and visual lobe under the up-and-down slope (direction) factors and the longitudinal slope (percent) factor overall were different. e interaction between the up-and-down slope (direction) factors and the longitudinal (percent) factor on the average fixation time and visual lobe were notable. e average fixation time driver under the uphill factor was smaller than the average fixation time under the downhill factor, whereas the visual lobe under the uphill factor was significantly larger than the visual lobe under the downhill factor. Furthermore, the visual load in the uphill section was smaller than that of in the downhill section.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Ethical Approval
is study was conducted in full compliance with the laws. e tests were approved by the relevant local authorities. Collecting eye tracking data were approved by participants.

Disclosure
All data in this study were for scientific research only, not for commercial use.

Conflicts of Interest
e authors declare that they have no conflicts of interest.