Impact of Road Central Greening Configuration on Driver Eye Movements: A Study Based on Real Vehicle Experiments

Abstract

Safe driving depends on drivers’ ability to rapidly and accurately process information in varying traffic conditions. The presence of central green landscapes on roads is a key factor in this context. However, there is a gap in current research, which tends to focus on qualitative aspects of landscape design while lacking quantitative data-driven analyses. In this study, we aim to address this gap by investigating the impact of road central greening configuration on the eye movements of young novice drivers, a population particularly sensitive to external environmental changes. Specifically, we explore the influence of central green landscapes on four visual parameters: driver gaze, saccade, blinking, and pupil behavior. Through real vehicle experiments conducted on different road sections, we collected visual feature data to comprehensively analyze the patterns of driver eye movements. Our findings reveal that the introduction of central green landscapes can exert cognitive pressure on drivers, leading to specific alterations in their visual parameters. These changes include dispersed gaze points, reduced effective gaze durations, increased gaze frequencies, extended saccade durations and angles, heightened blink durations and frequencies, and reduced pupil area. By shedding light on the intricate interplay between road central greenery and driver behavior, this study provides valuable insights for optimizing landscape design in transportation planning and enhancing road safety considerations.

1. Introduction

As China advances through the process of modernization, our transportation sector has experienced rapid development. However, the rate of traffic accidents has not only failed to decrease but has also shown an increasing trend. Traffic accidents fundamentally occur due to an imbalance among the elements of human behavior, vehicles, roads, and the environment. Therefore, in order to strengthen traffic safety management and reduce the occurrence of traffic accidents, it is crucial to harmonize various factors within the transportation system, thus enhancing the overall coordination of the transportation system [1,2,3].

Abundant research from both domestic and international sources indicates that the behavior of drivers is significantly influenced by the external traffic environment. The key to safe driving lies in a driver’s ability to rapidly process information within the road traffic environment within a certain time frame, with processing time largely contingent upon the complexity of the road environment’s information [4,5].

Hence, in the face of the severe challenges to traffic safety, identifying issues within the road environment and analyzing them is one of the critical tasks in regard to the current road system [6].

Road greening construction, as an integral component of the road environment, is currently a hot topic focusing on aspects such as ‘returning to nature,’ ‘sustainable development,’ and ‘ecology’. Many countries are attempting to introduce green elements into road space designs by constructing public greenways and tree-lined pedestrian streets, thereby achieving road greening and creating comfortable living environments [7,8].

Both domestic and international scholars have conducted case studies and found significant effects of greenery on traffic safety. For instance, Kathleen L. Wolf and her team focused on the accident rates on American roads and explored the positive aspects of road greening in terms of aesthetics, environmental benefits, and economics. However, they also discovered negative impacts of road greening on traffic safety. Consequently, they raised a series of questions regarding the advantages and disadvantages of road greening construction [9]. Hans Antonson and his colleagues studied three distinct types of roadside greenery in Switzerland through driving simulation experiments. They investigated how road greening influenced drivers’ psychological responses, driving speeds, and positioning. Their research underscored the importance of road greening for traffic safety. However, it should be noted that their study did not comprehensively consider other factors contributing to road safety hazards [10]. Tian Qing proposed a dialectical relationship between urban road greening and urban traffic safety, suggesting mutual influences and constraints. Based on this, they put forward designs for urban road segment greening and intersection greening projects with a focus on urban traffic safety. The research results highlighted the significance of factors such as plant height, planting spacing, plant selection scale, and color selection in road traffic safety [11]. Tang Guilan and her team utilized driver visual feature analysis methods and relevant specifications to establish models for straight and curved road segments. These models were used to determine the optimal line-of-sight induction results for drivers at different speeds, forming the basis for accurate determination of planting spacing for greenery [12].

Scholars worldwide have used eye-tracking systems to conduct research on driver fatigue detection, visual search, route indicators, and assessments of the rationality of traffic facilities. For example, British scholar T. Luke employed an eye tracker to monitor the eye movements of train drivers, collecting various eye movement parameter data and conducting in-depth research on the regularities of train driver gaze patterns [13]. Michelle L. Rusch and her team investigated the application of enhanced line-of-sight induction. Their results indicated that enhanced line-of-sight induction effectively assisted drivers in quickly detecting roadside hazards and reducing drivers’ response time to dangers, thereby significantly improving driving safety [14]. Yang Yunxing, based on the landscape characteristics of mountainous highways, classified green landscapes into three classic spatial design categories. They divided the driver’s gaze area into seven areas of interest. The study revealed that different spatial design categories led to different distributions of driver gaze points, which were reflected in data such as driver pupil diameter, gaze ratio, gaze duration ratio, and average gaze duration [14]. Yang Yunxing and his team conducted real vehicle tests to collect data on driver gaze span, gaze percentage, gaze duration, skin conductance, electrocardiography (ECG), and heart rate while driving in spaces with different levels of enclosure on highways. The results showed that skin conductance values were more effective in representing driver psychological states than ECG and heart rate. As enclosure levels increased on the highway, driver skin conductance and gaze span values significantly increased, while blink frequency and gaze duration decreased noticeably [15].

Considering the current state of both domestic and international research, it is clear that although studies have delved into the influence of green landscapes on traffic safety, particularly regarding drivers’ eye movement behavior, several unresolved issues remain.

(1)

In the realm of research content, it is essential to account for the varying legal regulations and cultural traditions across different countries. Furthermore, there are distinct differences in research concerning road landscapes between domestic and international studies. In contrast to developed countries, research in China primarily focuses on landscape design analysis, often remaining in the qualitative analysis stage, with limited quantitative analysis of green landscapes.

(2)

From the perspective of the research object, there is a dearth of classification research on drivers in studies related to road greening landscapes, both domestically and internationally. Considering factors such as age and driving experience, young novice drivers are more susceptible to the influence of changes in green landscapes on the road due to their age-related characteristics and limited driving experience. However, this specific group of drivers has received insufficient attention in current research.

(3)

From the perspective of research methods, most studies involve simulated driving experiments conducted in a driving simulator or cockpit, encompassing data collection and analysis during the driving process. However, it is crucial to acknowledge the substantial disparities between simulated environments and real driving conditions. Furthermore, participants’ psychological and physiological states may exhibit variations between simulated and actual driving environments.

In summary, conducting real vehicle driving tests in actual traffic environments to analyze the impact of road greenery settings on the eye movement behavior of young novice drivers will impart greater theoretical significance and research value to the results of the experiments. The outcomes of this research will also establish a theoretical foundation for road greening design and the advancement of the field of traffic safety studies.

In the following sections, we will delve into various aspects of this study. In Section 2 of the paper, we delve into the Materials and Methods used in our research. Moving on to Section 3, we present the outcomes of our experiments. In Section 4, we critically analyze and quantitatively evaluate the experimental results. Finally, in Section 5, we draw the paper to a close with a comprehensive summary and an outlook for future research.

2. Materials and Methods

2.1. Apparatus

(1)

Experimental Vehicle: In order to ensure the accuracy of the test results, a fixed test vehicle model was used for this experiment to prevent interference from factors such as vehicle size, operating condition, and driving characteristics, thus ensuring the reliability of the results. The chosen test vehicle was a Nissan Teana, a four-door sedan with a body length of 4.850 m and a width of 1.795 m.

(2)

Eye-tracking Equipment: The structure of the eye-tracking glasses is illustrated in Figure 1. The eye-tracking device utilized in this experiment was the Dikablis Glasses 3.0 eye-tracking glasses manufactured by Ergoneers, a German company. It boasts a pupil tracking accuracy of 0.1° and a sampling frequency of 60 Hz. The eye-tracking glasses are designed in the form of glasses, allowing for easy and unobtrusive wearing on the driver’s head without interfering with their driving activities. A full-scene high-definition camera is positioned at the center of the eye-tracking glasses, approximately at the wearer’s brow level, primarily aimed at capturing the driver’s central field of view image data. The eye camera is situated beneath the eye-tracking glasses, aligned with the driver’s eye, facilitating the real-time collection of eye movement data reflecting the driver’s current state.

2.2. Participants

For similar on-road experiments, the typical number of participants fell within the range of 4 to 6 individuals. Considering the specific driving routes and distances designed for this experiment, a total of 8 young novice drivers were selected as participants. The age range of the participants was between 18 and 26 years old, with driving experience ranging from 0 to 3 years. All participants were in good physical health and had no vision impairments. In the week preceding the experiment, participants were required to maintain good physical health, refrain from any medication use, maintain regular and stable daily routines, and ensure an adequate amount of sleep. Prior to the commencement of the experiment, participants were provided with detailed information about the experiment, and they voluntarily agreed to participate by signing an informed consent form.

2.3. Experimental Environment and Duration

For this experiment, it was essential to select a city road with multiple lanes and rich central green landscaping. After careful consideration, Jiangxi Province’s Zijin Avenue in Nanchang City was chosen as the experimental road segment. Zijin Avenue serves as a primary route connecting Nanchang City with the Han Dynasty Haihunhou Ancient Archaeological Site Park. This road segment is approximately 15 km in length, with a speed limit of 60 km/h. The road is designed with four lanes, divided into two lanes in each direction. It features diverse road landscaping, and the terrain on both sides of the road is relatively flat, with predominantly straight and long sections. This choice of road segment provides a suitable real-world environment for conducting the experiment, allowing for the evaluation of the impact of road greenery settings on the eye movement behavior of young novice drivers.

The experiment was conducted during the following time period: from 2 October 2022, to 5 October 2022, with testing sessions held in the morning from 9:00 a.m. to 11:00 a.m. and in the afternoon from 2:00 p.m. to 5:00 p.m. This specific date and time range was chosen due to several favorable factors. During these days and times, the traffic volume on the experimental road segment was relatively low, the temperature was moderate, lighting conditions were excellent with high visibility, and there were no significant factors affecting the participants. Additionally, the selected time slots ensured that the participants were in a good mental and physical state during the driving sessions.

2.4. Experimental Protocol

(1)

Preparation stage before experiment

This stage primarily involved three main tasks:

(i)

Preparation of experimental vehicles: before the experiment, a thorough inspection of the experimental vehicles was conducted. This included checking the vehicle’s overall condition, confirming the fuel tank load capacity, and ensuring clean windows that would not interfere with the driving process.

(ii)

Preparation and calibration of experimental equipment: cameras were installed on the front and rear of the vehicle, as well as on the left side of the driver. The functionality of the Dikablis Glasses 3.0 eye-tracking device was verified, and the portable power supply was checked to ensure sufficient usage time.

(iii)

Preparation and adaptation of participants: participants were fitted with the eye-tracking device, and the connection with the computer was verified. Participants underwent an adaptation phase where they familiarized themselves with the vehicle. During this period, participants’ physical condition was assessed, and they were queried about any potential interference caused by the equipment. The driving task and safety precautions were also explained to the participants.

(2)

Experiment execution stage

During this stage, participants were instructed to adhere to traffic regulations and drive naturally, maintaining their usual driving style. The following steps were taken:

(i)

Initiation of experiment: Participants were instructed to choose a lane adjacent to the central greenery of the road and maintain their position in that lane throughout the entire length of Zijin Avenue. Changing lanes or overtaking other vehicles was prohibited during the experiment. Completing the entire journey along Zijin Avenue constituted one complete trial. After finishing each trial, participants were advised to briefly pause at a safe location for rest to prevent driver fatigue from extended driving. Once participants were confirmed to be in good condition, they proceeded to the second trial.

(ii)

Second trial: In the second trial, participants were required to choose and stay in a different lane, this time away from the central greenery, while following the same driving requirements. Similarly to the first trial, driving the entire length of Zijin Avenue constituted the completion of all trials.

(3)

The end of the experiment

In this final stage, the vehicle was parked in a safe location, the engine was turned off, and the experimental data were checked for completeness and accuracy. The experimental equipment was turned off and assessed for its condition, and the power supply device’s remaining charge was evaluated.

2.5. Data Analysis

2.5.1. Segmentation of Experimental Roadway

In order to investigate the impact patterns of driver visual characteristics with or without the presence of central green landscapes, it was necessary to partition the test sections. Based on the actual road conditions in the real vehicle experiments, the test sections were divided into four categories [16,17]: far lane with central green landscapes, near lane with central green landscapes, far lane without central green landscapes, and near lane without central green landscapes, as illustrated in Figure 2.

2.5.2. Handling of Anomalous Data

In the course of the experiment, two common types of anomalous data were detected in the eye-tracking dataset.

The first type consisted of data values that significantly deviated from the normal range. For instance, instances where the driver’s gaze duration is substantially lower than the average gaze duration by 100 ms are considered anomalous. Typically, such data were handled by excluding them from the dataset.

The second type of anomalous data consisted of data points within the eye-tracking dataset that exhibited a noticeable offset from the overall sample data. These anomalies can profoundly affect the overall data variations and have an adverse impact on data analysis. The occurrence of the second type of anomalous data was inevitable due to the inherent interference of the monitoring equipment worn by the drivers in their driving behavior. The drivers’ driving states differed to some extent from their ideal states, and the vehicle could not maintain a consistent driving pattern throughout the entire course, resulting in fluctuations in the vehicle’s trajectory.

Choosing appropriate methods for handling anomalous data is crucial for effectively screening the dataset and enhancing data accuracy.

To address the presence of the second type of anomalous data mixed within the normal dataset, a commonly employed screening method is the Grubbs’ test, also known as the $3\sigma$ criterion [18]. The principle of this method is that for $x_i$ data points:

$$3\sigma \ge \left|x_i-\overline{X}\right|$$

(1)

where $x_i$: the i-th data point in the dataset; $\overline{X}$: the average of the dataset; $\sigma$: the standard deviation of the dataset.

The data $x_i$ are considered valid data; in this study, eye gaze data from five participants with significant differences in eye size were selected for the delineation of areas of interest. The coordinates of gaze points for these five participants during the driving process were extracted, and a gaze point distribution plot was generated. In the distribution plot, horizontal angles were represented on the X-axis, while vertical angles were represented on the Y-axis.

The distribution of gaze points before and after abnormal data processing is shown in Figure 3.

2.5.3. Gaze Area Division

During driving, drivers need to adjust their vehicle’s posture by visually gathering information about the surrounding road environment. In this process, the driver’s gaze points will pause at different targets. These targets are divided based on certain patterns, resulting in different gaze areas for drivers during the driving process.

The K-means clustering analysis method involves first determining the distance between each gaze point data point. Subsequently, it identifies parameters representing the similarity between each gaze point data point. Finally, based on these similarity parameters, it classifies these data points and determines the distribution of the driver’s gaze areas. The advantage of clustering analysis is its ability to effectively mitigate the impact of individual differences among drivers on the partitioning of gaze areas, thereby improving the accuracy of the results [19,20].

Basic steps of the K-means clustering algorithm:

Step 1: Choose any k data points from the experimental data as initial centroids.

Step 2: Calculate the distances between each data point and the initial centroids using the Euclidean distance method:

$$D(x_i,c_j)=\sqrt{{(x_i^1-C_j^1)}^2+\cdots +{(x_i^k-c_j^k)}^2+\cdots +{(x_i^D-c_j^D)}^2}$$

(2)

where $c_j$: the center of the j-th cluster; $D(x_i,c_j)$: the distance from $x_i$ to $c_j$; $x_i^k$ and $c_j^k$: the k-th dimensional coordinates from $x_i^k$ to $c_j^k$;

Since the eye-tracking data coordinates exist in a two-dimensional coordinate system, it is possible to further optimize the formula as follows:

$$D(x_i,c_j)=\sqrt{{(x_i^1-c_j^1)}^2+{(x_i^2-c_j^2)}^2}$$

(3)

Step 3: Calculate the distance from each data point to the initial centers, then compute their similarity. Assign each data point to the corresponding category of the nearest cluster center based on the nearest principle. The formula is as follows:

$$M(x_i,c_j)=\frac{1}{D(x_i,c_j)}$$

(4)

Step 4: Classify data points of the same type and identify new cluster centers. The coordinates of the new cluster centers are determined by the following formula:

$$c_j^k=\frac{x_{j1}{}^2+x_{j2}{}^2+\cdots +x_{j{N}_j}^2}{N_j}$$

(5)

where $x_{j1}{}^2$: the two-dimensional coordinates of the first point in the j-th class; $N_j$: the number of data points in the j-th class.

Step 5: Repeat Step 3 and Step 4 until the algorithm converges.

Based on the principles and steps of the K-means clustering algorithm described above, we implemented the algorithm using MATLAB. We edited the code to analyze eye-tracking data from participants after preprocessing it to handle outliers. We conducted the analysis with different cluster numbers, namely 4, 5, 6, and 7. As shown in Figure 4, it was found that when the clustering number was set to 4, it was closer to the real situation of the driver’s gaze area. A summary of cluster centers and sample counts is shown in

Table 1. Cluster center and sample count (when the number of clusters is 4).

Category	Initial Cluster Centers	Final Cluster Centers	Sample Count
1	(239.829, 187.038)	(233.68052, 153.02635)	8158
2	(294.156, 138.004)	(259.54992, 139.57205)	1602
3	(220.865, 110.874)	(236.46669, 136.53271)	17,992
4	(153.552, 158.98)	(193.44425, 143.28735)	1099

.

Additionally, combining the clustering results with the actual driving behavior patterns can provide a better interpretation of the clustering outcome. The driver’s areas of interest were divided into four parts: the area to the left of the vehicle, the area to the right of the vehicle, the area of the road ahead of the vehicle, and the area of the dashboard inside the vehicle. To analyze the impact of central green landscapes on driver behavior, the area to the left was considered equivalent to the central green landscape area. The division is illustrated in Figure 5.

3. Results

3.1. Gaze Characteristics Analysis

(1)

Gaze point distribution

The hot spot map offers a qualitative and intuitive representation of a driver’s field of vision and its dynamics across various road segments featuring central greenery landscapes. Examining the hot spot map provides valuable insights into driver gaze and saccade behaviors. Red areas represent regions of the highest visual attention, while yellow and green areas signify a gradual reduction in gaze intensity [21].

From the analysis of the results in Figure 6, it can be seen that the existence of the central green landscape leads to an increase in the dispersion of the driver’s gaze points. When vehicles drive on the lane with central green landscape in the distance, some attention points tend to focus on the greenery on both sides of the road. However, most drivers’ eyes are still focused on the road ahead. In contrast, when the vehicle is driving on a lane with a central green space nearby, the gaze points are significantly dispersed, and the proportion of gaze points focused on the road ahead is reduced. Drivers’ gaze points showed large deviations in both horizontal and vertical directions, suggesting that their visual recognition abilities may be disrupted. In the absence of a central green landscape, the driver’s gaze changes when transitioning between lanes. When driving on a driveway without a distant central green feature, there is minimal visual distraction, causing the driver to focus on the area directly ahead of the road. However, when the vehicle approaches the central green landscape area, the driver’s vision begins to diverge and turn to both sides of the road ahead.

(2)

Effective gaze duration and gaze frequency

Based on relevant literature, gaze behaviors with a duration greater than 100 milliseconds in the experimental data are considered effective gaze instances. Using this criterion, the average duration of effective gaze is calculated for drivers in different sections of the test route.

See Figure 7 for the statistical results of effective gaze duration and gaze frequency. Analysis reveals that the range of effective gaze duration for drivers in different sections of the test route varies between 300 and 800 milliseconds, indicating significant data variability. The longest gaze duration is observed in the section without central landscaping in the remote lane. This is attributed to the minimal impact of the road environment on drivers in this section, allowing them to focus their gaze on a specific point on the road for an extended period. Conversely, the shortest gaze duration is observed in the section with central landscaping in the nearby lane. This section features complex central landscaping, and drivers’ proximity to it results in shorter individual gaze durations. Gaze durations in this section predominantly concentrate on the road ahead and the central landscaping area to the left of the vehicle.

Based on the comprehensive analysis of the results, the following conclusions can be drawn: The gaze characteristics of drivers undergo significant changes across different road segments. In road segments without central green landscapes in the distance, drivers exhibit long-duration and low-frequency gaze behavior. However, as central green landscapes appear and approach, this gaze behavior gradually transforms into shorter-duration and higher-frequency gaze patterns. This transformation can be attributed to the presence of central green landscapes, which occupy a portion of the driver’s visual resources, increase the visual information objects that need to be processed, and consequently lead to a decrease in the duration of individual gazes and a significant increase in gaze frequency.

3.2. Saccade Behavior Analysis

(1)

Duration of saccade

Saccade duration refers to the duration of a single saccade behavior, measured in milliseconds. The saccade duration of drivers is primarily influenced by the complexity of the driving environment. As the complexity increases, drivers frequently shift their gaze between multiple targets to gather relevant information from their surroundings. This dispersion of attention away from the road ahead can lead to reduced focus and increased reaction times in response to sudden events, posing safety concerns during driving.

(2)

Angle of saccade

Saccade angle is a crucial parameter that reflects the range of eye movements between the end of one gaze and the beginning of another during a specific period. When analyzing drivers’ saccade behavior in different experimental road segments, absolute values of saccade angles are considered while disregarding their direction.

It can be seen from the statistical results in Figure 8b that when driving on a road section with no central green landscape in the far lane, the driver mainly pays attention to the area in front of the vehicle. From the results, it is evident that when driving on the road segment without central green landscaping in the far lane, drivers primarily focus on the area in front of their vehicle. Consequently, they only need to scan the road conditions within this specific area, resulting in smaller saccade angles. However, when driving on the road segment with central green landscaping in the near lane, drivers aim to gather more information about the road, particularly from the central green landscaping area on the left side. This leads to rapid shifts in drivers’ gaze points between the left-side central green landscaping area and the area in front of the vehicle. Consequently, in this experimental road segment, drivers exhibit larger and more concentrated saccade angles.

3.3. Blink Characteristic Analysis

(1)

Blink duration

Blink duration refers to the average time taken by a driver for each blink, encompassing the entire process from fully opening the eyes to closing and then reopening them.

Based on the analysis of the statistical results in Figure 9a, it is evident that when driving in lanes without central greenery in the far distance, drivers exhibited shorter and more fluctuating blink durations. This implies that drivers were in a relatively relaxed driving state, with lower cognitive pressure from the road environment. As central greenery started appearing and getting closer, blink durations showed a slight increase but fluctuations reduced, indicating that drivers unconsciously adjusted their blink durations to concentrate on observing the road environment. When drivers were in lanes with central greenery nearby, they faced the most complex road environment, and to alleviate the increased cognitive pressure, their blink durations increased accordingly.

(2)

Blink frequency

Blink frequency refers to the number of blinks per unit of time. Studies have found that when a driver’s attention is scattered, the blink frequency tends to increase. Therefore, blink frequency can serve as a reference indicator for assessing a driver’s cognitive response and visual ability.

Based on the analysis of the statistical results in Figure 9b, the following observations can be made: When driving on the far lane without central green landscape, the road environment is relatively uniform, resulting in lower cognitive stress for drivers, and thus, a lower blink frequency is observed. However, with the appearance and proximity of the central green landscape, the driver’s attention gradually disperses, leading to an increase in blink frequency. Particularly, when driving on the near lane with central green landscape, the complex and dynamically changing nature of the central green landscape occupies a significant portion of the driver’s visual resources, resulting in increased cognitive pressure and a noticeable rise in blink frequency.

3.4. Pupil Characteristic Analysis

Pupil area: The size of the pupil is regulated by the contraction of muscles within the iris, a process controlled by the sympathetic nervous system in the eye. Drivers can use their pupils to control the amount of light entering their eyes, adapting to varying environmental conditions. Additionally, pupil size is influenced by factors such as attention and physiological state.

Pupil area is a crucial indicator for assessing a driver’s cognitive activities, resource allocation, and processing capacity. Pupil area varies with the intensity of visual information processing, reflecting the driver’s current level of stress and visual workload. When drivers are stressed, their pupils tend to constrict. The greater the visual workload, the more pronounced the changes in pupil size. Pupil area data from drivers in different experimental road segments were collected and analyzed.

The analysis of the results based on Figure 10 reveals that when driving in the far lane of the road without central greenery, drivers exhibit relatively large but stable pupil areas, indicating a relaxed driving state. As central greenery comes into view, driver pupil areas decrease, signifying that the visual information from the road environment is starting to exert some pressure on the drivers’ vision. When drivers are in the lane with central greenery nearby, their pupil areas become smaller and show greater fluctuations. This suggests that the road environment is more complex, putting drivers in a tense driving state. In this situation, drivers need to continuously shift their focus between the left-central greenery area and the front-road area due to increased information in the road environment. This elevated visual workload adds to the drivers’ visual stress, resulting in noticeable fluctuations in pupil area.

4. Discussion

4.1. Quantitative Discussion of Driver Eye Movement Behavior

The analysis of gaze characteristics, saccadic behavior, blink characteristics, and pupil characteristics across various road segments with and without central greenery landscapes has yielded quantitative insights into how road landscapes influence driver eye movements.

4.1.1. Gaze Characteristics

The hot spot map analysis provides valuable quantitative insights into the distribution of driver gaze points. It is evident that the presence of central greenery landscapes leads to increased dispersion in drivers’ gaze points. In road segments with distant central greenery landscapes, drivers maintain their focus primarily on the road ahead, with only approximately 8–10% of gaze points directed towards the surrounding greenery. However, in lanes with nearby central greenery landscapes, there is a significant dispersion of gaze points, with up to 25–30% of gaze points directed away from the road ahead. This dispersion suggests a notable shift in drivers’ visual attention away from the road ahead, indicating potential interference with their visual recognition abilities. These quantitative findings underscore the substantial impact of central greenery landscapes on driver gaze distribution.

4.1.2. Saccadic Behavior

The analysis of saccade behavior has quantitatively revealed the role of the driving environment in determining the duration and angle of saccades. In road segments without central greenery landscapes in the distance, drivers exhibit smaller saccade angles, typically ranging from 2.5 to 5.0 degrees, as they primarily focus on the area in front of their vehicle. In contrast, when central greenery landscapes are present, drivers tend to exhibit larger saccade angles, ranging from 4.5 to 6.0 degrees. This suggests that the introduction of central greenery landscapes significantly increases the cognitive demands on drivers as they navigate visually more complex environments, with saccades becoming more frequent and extensive.

The analysis of saccade behavior revealed that the complexity of the driving environment plays a pivotal role in determining the duration and angle of saccade. In road segments without central greenery landscapes in the distance, drivers exhibit smaller saccade angles as they primarily focus on the area in front of their vehicle. In contrast, when central greenery landscapes are present, drivers tend to shift their gaze more rapidly between the central greenery area on the left and the road ahead, leading to larger and more concentrated saccade angles. These observations suggest that the introduction of central greenery landscapes increases the cognitive demands on drivers as they navigate more visually complex environments.

4.1.3. Blink Characteristics

The analysis of blink characteristics has provided specific insights into the impact of central greenery landscapes on blink duration and frequency. When driving in lanes without distant central greenery, drivers exhibited relatively short and fluctuating blink durations, typically lasting 150–200 ms. This indicates a relatively relaxed driving state. However, as central greenery appeared and approached, drivers exhibited longer blink durations, with durations extending to 250–300 ms, and reduced fluctuations, reflecting their conscious effort to concentrate on observing the evolving road environment. In lanes with nearby central greenery landscapes, blink frequencies increased significantly, with drivers blinking 25–30 times per minute, indicating the elevated cognitive pressure induced by the visually complex and dynamically changing central greenery.

4.1.4. Pupil Characteristics

Analysis of pupil characteristics further supports the impact of road landscapes on driver cognitive activities. When driving in lanes without central greenery in the far distance, drivers exhibited relatively large but stable pupil sizes, indicating a relaxed driving state. However, as central greenery came into view, pupil sizes decreased, suggesting the onset of visual-information-related stress. In lanes with central greenery nearby, drivers exhibited smaller pupil sizes with greater fluctuations, indicating heightened visual workload and stress. The continuous shifts in visual focus between the central greenery area and the road ahead contributed to these fluctuations. These findings emphasize the potential of pupil characteristics as indicators of driver cognitive load.

The analysis of pupil characteristics has quantitatively reinforced the impact of road landscapes on driver cognitive activities. When driving in lanes without central greenery in the far distance, drivers exhibited relatively large but stable pupil sizes, typically resulting in a pupil area of approximately 1550–1700 square pixels, indicating a relaxed driving state. However, as central greenery came into view, pupil sizes decreased to an area of around 1300–1400 square pixels, suggesting the onset of visual information-related stress. In lanes with central greenery nearby, drivers exhibited smaller pupil sizes, ranging from 1250 to 1450 square pixels, with greater fluctuations, indicating heightened visual workload and stress. These fluctuations were particularly pronounced when drivers shifted their focus between the central greenery area and the road ahead.

4.2. Driver Eye Movement Behavior Disparities: Young vs. Experienced

To gain a deeper understanding of the cognitive differences among drivers of varying ages and experience levels when navigating central greenery, this study conducted an analysis of the eye movements of young and experienced drivers through literature review. Initially, young drivers demonstrate higher cognitive abilities and reaction speeds, enabling them to make driving decisions rapidly. However, they are more susceptible to distractions and exhibit higher levels of risk-taking behavior. In contrast, experienced drivers possess richer driving experience, allowing them to more effectively navigate complex traffic scenarios, displaying a prudent and cautious driving style with a focus on actual road conditions and safety factors.

Regarding the study results, under similar road conditions, experienced drivers may exhibit distinct eye movement behaviors. In terms of gaze characteristics, they are more likely to maintain concentrated attention on the road, even in the presence of central greenery, resulting in a lower degree of gaze dispersion. In terms of scanning behavior, they demonstrate greater stability, as they are better adapted to processing intricate road landscapes. Additionally, the blink frequency of experienced drivers may slightly increase in similar situations, albeit to a lesser extent than observed in young drivers. Moreover, their pupils consistently maintain a relatively lower cognitive load state.

4.3. Comparative Analysis with Previous Studies

This study seeks to address the research gap regarding the influence of road center greening on driver eye movement behavior. In contrast to prior qualitative investigations, this paper presents a more quantitative analysis facilitated by actual vehicle experiments and the collection of data on driver visual characteristics. These approaches enable a precise evaluation of the genuine impact of central greenery on driver cognitive load. The utilization of real vehicle experimental methods enhances ecological validity, providing a more realistic simulation of actual driving conditions compared with alternative approaches such as simulated driving. The research findings underscore the pronounced influence of central greening on eye movement behavior, furnishing practical data support for road design and safety planning. Distinguished from preceding research, this paper places a heightened emphasis on practical applications, contributing to advancements in driver safety and comfort.

5. Conclusions

In this study, we focused on the influence of central green landscape features on the driving behavior of young novice drivers. Through real-world driving experiments and eye-tracking data analysis, we draw the following conclusions:

(1)

Road Segment Classification and Data Processing: We successfully categorized the experimental road segments based on the presence of central greenery and lane configurations. Employing the Lylida criterion and K-means clustering, we effectively processed and analyzed eye-tracking data, defining four distinct regions of interest for further investigation.

(2)

Changes in Visual Dimensions: Exploring four key visual dimensions—gaze, saccades, blink frequency, and pupil characteristics—we observed that the introduction and proximity of central green landscapes significantly influenced drivers’ eye movements. In such situations, drivers exhibited reduced effective gaze duration, increased gaze frequency, extended saccade duration and angle, higher blink frequency, slightly prolonged blink duration, and reduced pupil size.

(3)

Increased Cognitive Load on Drivers: Based on the observed changes in eye-tracking data, we conclude that the presence and proximity of central green landscapes impose a notable cognitive burden on drivers, leading to heightened cognitive stress and manifesting as a more vigilant and tense driving state.

However, this study also has some limitations and areas for improvement, which warrant further research and refinement:

(1)

Sample Size: In future studies, expanding the sample size would help to provide a more comprehensive understanding of the impact of central green landscapes on different types of drivers and validate the research findings on a wider scale.

(2)

Data Collection: Future research can collect more detailed data related to driver behavior and visual characteristics to more accurately quantify changes in cognitive load and attention allocation.

In summary, this study provides robust empirical support for understanding the impact of road landscape, particularly central green features, on drivers’ eye movements and driving behavior. Future research will address these limitations to obtain deeper research findings on the role and effects of road greenery in traffic safety.