This work is licensed under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
1. Introduction
Young drivers are getting a lot of attention because they are involved in more road traffic accidents. Young male drivers have lower hazard perception (HP) skills than older and more experienced drivers, but there is a tendency for them to overestimate their driving skills in hazardous situations, and both factors contribute to an overrepresentation in traffic accidents [1]. Hazard perception is usually considered as the ability to “read the road” or the awareness of hazardous situations [2]. It is the process of detecting, evaluating, and responding to dangerous events on the road that have a high likelihood of leading to a collision [3, 4]. Young experienced drivers reacted to covert hazards and overt hazards faster than young novice drivers [5]. Young drivers’ objective and subjective HP skills were consistent, although only for visible hazards. Also, drivers who responded in time had significantly higher subjectively assessed hazard perception and driving skills than drivers who did not respond to the hazards [1, 6]. Meanwhile, eye movement dynamics and change in pupil diameter can provide good measures of the drivers’ hazard perception and prediction [7]. Differences in eye movement patterns are often found when comparing passive viewing paradigms to actively engage in everyday tasks. Arguably, investigations into visuomotor control should therefore be most useful when conducted in settings that incorporate the intrinsic link between vision and action. The interactivity of simulated driving increases the demands on visual and attention than simply viewing driving movies [8]. Subtasks (e.g., letter search) can interfere with hazard identification and the allocation of attention [9, 10].
The information relevant to driving is predominantly visual [11]. The introduction of eye tracking technology provides a measure to record the location of eye fixation and saccade duration as well as visual search strategies [12, 13]. Many studies based on eye movement have revealed some interesting and instructive findings. For example, some hazards (e.g., behavioral prediction hazards, environmental prediction hazards, and dividing and focusing attention hazards) are more attractive than others to be easily fixated on [13], and novice or inexperienced drivers do not have the optimum visual search strategies compared to experienced drivers [3, 14–20]. It is worth noting that most works on drivers’ eye movements are focused on the visual characteristics in specific traffic environments or situations [4, 21–23]. However, what we can read from the performance of eye movements is not attached more attention, although relevant research studies are in the direction of drivers’ intent inference via eye behavior analysis [24]. In fact, drivers’ eye movement can possibly tell us more information, such as the situation they are facing, the level of mental load, visual behaviors for the traffic scenes, necessary information for a decision, and so on. To accomplish the task, it is necessary for us to discover the corresponding relationship between indices of eye movements and hazard perception.
Hazard perception tests are being used with greater frequency for driver training, assessment, and licensure [10], and they have become part of the graduated driver licensing (GDL) systems in countries, such as Australia, the UK, and the Netherlands [25]. In research studies related to hazard perception, four key methodologies (video, static image, simulator, and real-world test) were used for testing either singularly or in conjunction with other methods. The most used methodology was video, and the majority were filmed from the perspective of road users [26–30]. Static image tests were less commonly used and only studied for car drivers [31–33]. Static image tests may not be representative of the dynamic nature of real-world driving. However, both methodologies were correlated with self-reported driver errors and were able to discriminate between drivers based on age and experience [15, 16]. Also, there was no correlation between static image and video tests, suggesting that these methodologies may be tapping into different aspects of hazard perception [14]. Static image tests can reduce testing time and allow a larger variety of hazardous scenarios to be tested. They also have the benefit of explicit response time that can be calculated from stimulus onset while videos contain subjectively predetermined hazard windows that may fail to capture early responses [16]. In still images, it is often easier to select the moment of onset of one unambiguous hazard because hazard onset is synchronous with image onset in a static image, whereas an extended dynamic scene may create problematic response variability. For dynamic scenes, there is no uncertainty about determining the onset of a hazard, which is a factor that can complicate the calculation of response latency [8, 9, 33]. Although simulators may more accurately reflect the greater cognitive load experienced while driving [8] and can better emulate the on-road situations [13, 34, 35], the use of simulators is largely restricted to the costs associated with the setup. Real-world test is the most naturalistic method, but it is costly, time-consuming, and poorly controllable for hazards [16].
In the present study, to further search for the correlation hidden in indices of eye movements and hazard perception, we used static images taken from a young driver’s perspective to collect data of participants’ visual behaviors, subjective feeling/cognition, and objective response while inspecting the traffic scenes. The rest of the study is organized as follows. Section 2 further introduces the experimental method, gives the participant characteristics, materials, and equipment, and reports the content and procedures of the experiment. Section 3 reports the results of the experiment, the correlation analysis of participants’ hazard perception, the difference analysis of the participants’ eye movements, and the correlation between hazard perception and eye movements. Then, the results are discussed in Section 4. Finally, Section 5 reports the conclusion of the study.
2. Method
2.1. Participants
A total of 46 young drivers participated in this study. All participants held a valid driver license and had normal or corrected-to-normal vision. Every participant had a good willingness to participate. They were invited to fill out a demographic questionnaire, which included age, gender, driving experience, driving mileage, number of collisions involved, and traffic citation in the previous 3 years, as shown in Table 1. Table 1 shows the description and coding of drivers’ demographic information.
Table 1
The information of participants.
Drivers’ demographic information | Coding | Number | Proportion (%) |
Age | 1⟶age of 21–25 years | 31 | 67.391 |
2⟶age of 26–30 years | 15 | 32.609 | |
Gender | 1⟶male | 34 | 73.913 |
2⟶female | 12 | 26.087 | |
Driving experience | 1⟶less than 1 year | 11 | 23.913 |
2⟶1–3 years | 19 | 41.304 | |
3⟶more than 3 years | 16 | 34.783 | |
Driving mileage | 1⟶less than 5,000 km | 19 | 41.304 |
2⟶5,000–10,000 km | 18 | 39.130 | |
3⟶10,000–50,000 km | 7 | 15.217 | |
4⟶more than 50,000 km | 2 | 4.348 | |
Collisions involved in the previous 3 years | 1⟶none | 43 | 93.478 |
1⟶no less than 1 | 3 | 6.522 | |
Traffic citations in the previous 3 years | 1⟶none | 37 | 80.435 |
2⟶no less than 1 | 9 | 19.565 |
2.2. Materials and Apparatus
Considering the feasibility and benefits of using static scenes [8, 9, 15, 16, 33], a total of 35 traffic scenes containing hazards were used to test participants, which were taken from a driver’s perspective in a car traveling along different road situations. In order to analyze the possible influence of the traffic characteristics on the driver's hazard perception and eye movement, the scenes fell into five categories according to the similarity in traffic environment or hazard instigators. The reference characteristics for dividing traffic scenes are as follows.
Category 1 (vehicle lane-changing or braking on road): the category is characterized by vehicles ahead braking or changing into the lane that the instrumented car was in. This category contains 12 traffic scenes. Category 2 (vehicle turning at junctions or merging at ramps): the scenes show that when the instrumented car is approaching junctions or ramps, other cars are turning or merging and possibly invaded or had invaded the lane of the car. This category involves 5 traffic scenes. Category 3 (cyclist appearing from junction/ramp or traveling on lane/roadside): this category contains 7 traffic scenes. Category 4 (pedestrian crossing the road or waiting to cross): there are 6 traffic scenes in this category. Category 5: there are hazards which were invisible but possible to occur. The category involves 5 traffic scenes. As examples, five typical scenes are shown in Table 2.
Table 2
Examples of typical traffic scenes.
Categories | Typical scenes |
Category 1 | |
Category 2 | |
Category 3 | |
Category 4 | |
Category 5 |
A 15.6-inch flat-screen monitor with a resolution of 1280 × 1024 was used to display traffic scenes. Participants sat directly facing the screen and holding a computer mouse to click relative regions where hazards were in. The click time was recorded with a hazard perception test software developed based on MATLAB GUI and Win10 system. The software was run on a computer with 4 GB RAM and Intel Core i53210M processor. At the same time, Dikablis head-mounted eye tracker and D-Lab software were used to record eye movement data. The data were recorded by another computer. As shown in Figure 1, Figure 1(a) shows the equipment involved in the eye tracker. After being connected and worn, it constitutes the situation shown in Figure 1(b). Figure 1(c) shows an eye tracker head-mounted device, which has three cameras on it, and the two on the bottom are used to collect eye movement data. The upper front camera is used to capture the front view of the subject, as shown in Figure 1(d).
[figures omitted; refer to PDF]
2.3. Test Content
The test content consisted of hazard perception and eye movement. We wanted to know participants’ subjective feeling/cognition, objective reaction, and corresponding visual behaviors during inspection of the scenes.
Hazard perception is a comprehensive concept [10, 14, 33], and indices of complexity degree evaluation (CDE) and hazardous degree evaluation (HDE) on traffic scenes [14, 33], hazard identification time (HIT) [10], number of hazards slowly identified (NHSI), and number of hazards unidentified (NHU) were brought into the analysis of participants’ hazard perception. Complexity degree and hazardous degree were measured by values ranging from 1 to 5, and bigger value indicated higher complexity or hazardous degree. Hazard identification time was the interval between scene onset and the first click on the hazard region. The number of hazards slowly identified was the number of participants whose hazard identification time exceeded definition identification time, which was set as 4 s in the study [9, 33]. The number of hazards unidentified (NHU) indicated the number of participants who did not identify hazard for a traffic scene. The data of HDE, CDE, and HIT were recorded by the hazard perception test software, and the data of NHSI and NHU were processed according to the hazard identification result.
Eye movement indices included horizontal eye activity (HEA), vertical eye activity (VEA), mean fixation duration (MFD), average number of fixations (ANF), mean saccade duration (MSD), mean saccade angle (MSA), and average number of saccades (ANS). Eye movement data were directly recorded by the D-Lab software while participants inspected the traffic scenes, and the related indicators above were analyzed subsequently with the help of the software.
2.4. Procedure
As shown in Figure 2, after recruiting participants, the researchers explained the study in detail and invited them to complete the demographic questionnaire shown in Table 1. Before the actual experimental test, 3 to 6 illustrative scenes (i.e., the scenes not included in the 35 traffic scenes) were used to practice to ensure that participants understood the testing procedure.
[figure omitted; refer to PDF]
The first task for participants was to read 35 traffic scenes displayed in the monitor while wearing the Dikablis head-mounted eye tracker, identify hazards in traffic scenes, and click the hazard region. Participants were fitted with the eye tracking equipment and calibrated to ensure the eye movement data can be collected. Traffic scenes were switched by participants. After finishing the test, participants were asked to evaluate the complexity degree and hazardous degree of each traffic scene with a corresponding value.
3. Results
3.1. Hazard Perception
The average values of the hazard perception indices are shown in Table 3. It is obvious that the number of hazards unidentified in Category 5 is much more than that in other traffic scenes. The potential hazards with no obvious cue in the scenes are much less likely to be identified.
Table 3
The average value of hazard perception indices.
Indices | CDE | HDE | HIT/s | NHSI | NHU | |
Category of traffic scenes | 1 | 2.741 | 2.348 | 2.052 | 2.833 | 4.333 |
2 | 2.247 | 1.884 | 2.442 | 3.000 | 5.200 | |
3 | 1.790 | 1.418 | 2.326 | 3.857 | 6.000 | |
4 | 2.865 | 2.349 | 2.550 | 6.333 | 5.000 | |
5 | 1.877 | 1.223 | 2.626 | 5.800 | 17.200 |
Note. CDE = complexity degree evaluation; HDE = hazardous degree evaluation; HIT = hazard identification time; NHSI = number of hazards slowly identified; NHU = number of hazards unidentified.
Analysis of variance (ANOVA) was used to compare the hazard perception indices across the categories of traffic scenes. As shown in Table 4, the hazardous degree evaluation (HDE), hazard identification time (HIT), and number of hazards unidentified (NHU) show significant difference between different categories of traffic scenes, respectively.
Table 4
ANOVA of participants’ hazard perception indices.
Indices | Sum of squares | df | Mean square | F | Sig. | |
CDE | Between groups | 6.768 | 4 | 1.692 | 2.242 | 0.088 |
Within groups | 22.636 | 30 | 0.755 | |||
Total | 29.404 | 34 | ||||
HDE | Between groups | 7.487 | 4 | 1.872 | 2.749 | 0.046 |
Within groups | 20.423 | 30 | 0.681 | |||
Total | 27.910 | 34 | ||||
HIT | Between groups | 1.719 | 4 | 0.430 | 7.161 | 0.000 |
Within groups | 1.801 | 30 | 0.060 | |||
Total | 3.520 | 34 | ||||
NHSI | Between groups | 70.086 | 4 | 17.521 | 1.873 | 0.141 |
Within groups | 280.657 | 30 | 9.355 | |||
Total | 350.743 | 34 | ||||
NHU | Between groups | 650.419 | 4 | 162.605 | 7.388 | 0.000 |
Within groups | 660.267 | 30 | 22.009 | |||
Total | 1310.686 | 34 |
Note. CDE = complexity degree evaluation; HDE = hazardous degree evaluation; HIT = hazard identification time; NHSI = number of hazards slowly identified; NHU = number of hazards unidentified.
Moreover, the post hoc tests (i.e., multiple comparisons) based on Bonferroni indicated that no statistically significant difference was found between any specific two categories of traffic scenes regarding the HDE. The hazard identification time (HIT) in Category 1 was much less than that in Categories 4 and 5. The number of hazards unidentified (NHU) in Category 5 was much more than that in Categories 1, 2, 3, and 4. Further, to test whether there was correlation between hazard perception indices, Spearman’s rank correlation coefficient was used in the analysis.
As shown in Table 5, a significant positive correlation (
Table 5
Correlations between the indices of hazard perception.
Indices | CDE | HDE | HIT | NHSI | NHU | |
CDE | Correlation coefficient | 1.000 | ||||
Sig. (2-tailed) | — | |||||
HDE | Correlation coefficient | 0.933 | 1.000 | |||
Sig. (2-tailed) | 0.000 | — | ||||
HIT | Correlation coefficient | −0.363 | −0.428 | 1.000 | ||
Sig. (2-tailed) | 0.032 | 0.001 | — | |||
NHSI | Correlation coefficient | −0.156 | −0.208 | 0.605 | 1.000 | |
Sig. (2-tailed) | 0.369 | 0.230 | 0.000 | — | ||
NHU | Correlation coefficient | −0.484 | −0.625 | 0.424 | 0.488 | 1.000 |
Sig. (2-tailed) | 0.003 | 0.000 | 0.011 | 0.003 | — |
Note. CDE = complexity degree evaluation; HDE = hazardous degree evaluation; HIT = hazard identification time; NHSI = number of hazards slowly identified; NHU = number of hazards unidentified.
At the same time, the correlation between driver characteristics and the above five indicators was also analyzed. With the help of ANOVA, the results suggested that the young driver characteristics in Table 1 have no significant impact on these indicators.
3.2. Eye Movement
Table 6 shows the average value of participants’ eye movement data in different categories of traffic scenes. Furthermore, as shown in Table 7, ANOVA results of participants’ eye movement indicate that no statistically significant difference is found between different categories of traffic scenes regarding indices of the horizontal eye activity (HEA), vertical eye activity (VEA), mean fixation duration (MFD), mean saccade duration (MSD), and mean saccade angle (MSA), but the average number of fixations (ANF) and the average number of saccades (ANS) represent significant difference between groups of traffic scenes (
Table 6
The average value of participants’ eye movement indices.
Indices | HEA/pixel | VEA/pixel | MFD/ms | ANF | MSD/ms | MSA/degree | ANS | |
Category of traffic scenes | 1 | 514.505 | 152.898 | 733.865 | 3.938 | 31.691 | 4.660 | 3.204 |
2 | 593.855 | 209.921 | 714.738 | 4.747 | 62.077 | 4.764 | 3.859 | |
3 | 604.934 | 170.185 | 740.101 | 4.652 | 57.809 | 4.776 | 3.499 | |
4 | 751.174 | 216.770 | 794.491 | 5.885 | 38.141 | 4.705 | 4.596 | |
5 | 1259.756 | 472.827 | 723.132 | 5.948 | 57.389 | 4.502 | 4.708 |
Note. HEA = horizontal eye activity; VEA = vertical eye activity; MFD = mean fixation duration; ANF = average number of fixations; MSD = mean saccade duration; MSA = mean saccade angle; ANS = average number of saccades.
Table 7
ANOVA results of participants’ eye movement indices.
Indices | Sum of squares | df | Mean square | F | Sig. | |
HEA | Between groups | 2111986.140 | 4 | 527996.535 | 0.681 | 0.611 |
Within groups | 23273379.402 | 30 | 775779.313 | |||
Total | 25385365.542 | 34 | ||||
VEA | Between groups | 391533.844 | 4 | 97883.461 | 1.158 | 0.349 |
Within groups | 2534888.939 | 30 | 84496.298 | |||
Total | 2926422.782 | 34 | ||||
MFD | Between groups | 22827.252 | 4 | 5706.813 | 0.135 | 0.968 |
Within groups | 1270993.702 | 30 | 42366.457 | |||
Total | 1293820.954 | 34 | ||||
ANF | Between groups | 22.722 | 4 | 5.680 | 3.587 | 0.017 |
Within groups | 47.509 | 30 | 1.584 | |||
Total | 70.231 | 34 | ||||
MSD | Between groups | 5744.399 | 4 | 1436.100 | 2.583 | 0.057 |
Within groups | 16681.101 | 30 | 556.037 | |||
Total | 22425.500 | 34 | ||||
MSA | Between groups | .267 | 4 | 0.067 | 0.631 | 0.644 |
Within groups | 3.176 | 30 | 0.106 | |||
Total | 3.443 | 34 | ||||
ANS | Between groups | 12.831 | 4 | 3.208 | 2.919 | 0.038 |
Within groups | 32.969 | 30 | 1.099 | |||
Total | 45.801 | 34 |
Note. HEA = horizontal eye activity; VEA = vertical eye activity; MFD = mean fixation duration; ANF = average number of fixations; MSD = mean saccade duration; MSA = mean saccade angle; ANS = average number of saccades.
In addition, the correlation between driver characteristics and the above eye movement indicators has been analyzed. The ANOVA results show that the mean saccade duration (MSD) of young drivers in different age groups is different, and the older the age, the shorter the saccade time (
3.3. Correlation between Hazard Perception and Eye Movement
To test whether there was inherent correlation between participants’ hazard perception and eye movement, we used the Spearman statistical method to analyze the data. The correlation analysis results shown in Table 8 indicate that the mean fixation duration (MFD) and mean saccade angle (MSA) are negatively related to the complexity degree evaluation (CDE) (
Table 8
Correlations between the indices of hazard perception and eye movement.
Indices | HEA | VEA | MFD | ANF | MSD | MSA | ANS | |
CDE | Correlation coefficient | 0.220 | 0.259 | −0.337 | −0.022 | −0.309 | −0.410 | −0.055 |
Sig. (2-tailed) | 0.203 | 0.133 | 0.048 | 0.901 | 0.071 | 0.014 | 0.753 | |
HDE | Correlation coefficient | 0.176 | 0.194 | −0.304 | −0.128 | −0.388 | −0.144 | −0.128 |
Sig. (2-tailed) | 0.313 | 0.264 | 0.076 | 0.464 | 0.021 | 0.409 | 0.464 | |
HIT | Correlation coefficient | 0.006 | −0.022 | 0.205 | 0.650 | 0.138 | 0.167 | 0.702 |
Sig. (2-tailed) | 0.972 | 0.900 | 0.238 | 0.000 | 0.431 | 0.339 | 0.000 | |
NHSI | Correlation coefficient | 0.268 | 0.280 | 0.069 | 0.705 | −0.140 | −0.027 | 0.751 |
Sig. (2-tailed) | 0.120 | 0.104 | 0.696 | 0.000 | 0.424 | 0.877 | 0.000 | |
NHU | Correlation coefficient | 0.043 | 0.067 | 0.276 | 0.269 | 0.120 | −0.225 | 0.299 |
Sig. (2-tailed) | 0.805 | 0.701 | 0.109 | 0.118 | 0.491 | 0.194 | 0.081 |
Note. HEA = horizontal eye activity; VEA = vertical eye activity; MFD = mean fixation duration; ANF = average number of fixations; MSD = mean saccade duration; MSA = mean saccade angle; ANS = average number of saccades; CDE = complexity degree evaluation; HDE = hazardous degree evaluation; HIT = hazard identification time; NHSI = number of hazards slowly identified; NHU = number of hazards unidentified.
4. Discussion
The study sought to investigate the correlations between hazard perception (HP) and eye movement, that is, to investigate the young drivers’ subjective feeling/cognition, objective response, and eye movement in specific traffic scenes and the mutual correlation between the three factors. The traffic scenes in the form of static images were taken from young drivers’ perspective. The static image-based approach can rule out factors of road users’ speed and sound (e.g., background noise and traffic sound), which possibly can affect participants’ eye movement. Previous research studies have investigated different groups of young drivers’ eye movement or hazard detection in different traffic scenes [8, 26, 27, 31, 36]. The objective and subjective HP skills of young drivers are consistent [1, 5–7], which is also proven in this experiment. However, the relationship between hazard perception and specific indicators of eye movement has not yet been distinguished. Our research is to seek the influencing relationship and possible causes. According to the above findings from Tables 5 and 8, all significant correlations between indices of hazard perception and eye movement are shown in Figure 3.
[figure omitted; refer to PDF]
Different categories of traffic scenes resulted in different hazardous degrees, hazard identification times, and number of hazards unidentified (
There is no significant difference in the horizontal and vertical eye movements between different traffic scene groups (
The above results and analyses reflect the possible correlations between young drivers’ hazard perception and eye movements, and the correlations seem to be related to the traffic scenes. This possibility has also been verified. Specifically, higher complex traffic scenes resulted in smaller saccade angle (
In addition, the results in Figure 3 show the correlation between the subjective and objective indicators of young drivers. We can speculate the subjective cognition of them through objective measurement indicators, and vice versa. For example, for the relevance of hazardous degree evaluation to hazard identification time, we can predict the driver’s subjective perception of a certain traffic scene (e.g., self-perception of the hazardous degree) through the driver’s hazard identification efficiency. At the same time, in terms of driver characteristics, for drivers with more driving mileage, the range of their horizontal eye activity is greater, which means that the more experience the drivers had, the wider the scope of their attention is and the stronger their awareness of hazard search is. The relatively shorter saccade time of older young drivers may also be due to the influence of driving experience [15, 18].
It should be noted that the test conditions in laboratory are not consistent with the real-world environment, which possibly resulted in different psychological feeling and visual behaviors of participants, but we can achieve the same controlled condition for all participants [38], and it is helpful to discover the inherent correlation between hazard perception and eye movement. In fact, we indeed have demonstrated some obvious findings in the study. In terms of future research directions, it is important to determine whether these findings are replicated in on-road testing. Future research should also examine correlations between the evolution procedures of traffic scenes, awareness, and eye movement, especially whether the fixations coincide with the possible traffic conflicts and its characteristics. According to drivers’ eye movement when driving in a real-world environment, maybe we can know more information about drivers with the development of further research on the correlations between eye movements, hazard perception, and traffic scenes.
5. Conclusion
In order to analyze the correlation between drivers’ hazard perception and eye movement, the study regarded young drivers as the object and implemented a static image-based hazard perception test with the help of different scenes and eye movement data collection equipment. At the same time, the difference analysis and correlation analysis were carried out based on the collected data. The obtained results suggest that in different traffic scenes, some indicators of driver’s hazard perception and eye movement reflect significant differences. Also, some indicators of hazard perception and visual behavior are closely related.
The research on hazard perception and eye movement can guide road design, traffic management, driver training, and the test of driver’s hazard perception ability. In order to avoid traffic accidents as much as possible, we can add auxiliary massages on the road or in the car for hazardous scenes that are easily overlooked. To reduce risky driving behaviors and the phenomenon of non-motor vehicles or pedestrians appearing on motor vehicle lanes, strengthening traffic management and creating a sound traffic safety climate are recommended. Some typical scenes can also be used to observe, analyze, and improve drivers’ hazard perception skill. The results of this paper are based on the analyses of the acquired data. The next step of the research can use more abundant samples, indicators, traffic scenes, and test methods to further analyze the correlation between drivers’ hazard perception and eye movement.
Acknowledgments
This study was supported by the Shaanxi Natural Science Basic Research Project (grant no. 2019JM-309), Fundamental Research Funds for the Central Universities (grant no. 300102228205), and National Natural Science Foundation of China (grant no. 51108040).
[1] L. Ābele, S. Haustein, M. Møller, L. M. Martinussen, "Consistency between subjectively and objectively measured hazard perception skills among young male drivers," Accident Analysis & Prevention, vol. 118, pp. 214-220, DOI: 10.1016/j.aap.2018.02.022, 2018.
[2] M. S. Horswill, F. P. McKenna, A Cognitive Approach to Situation Awareness: Theory and Application, pp. 155-175, 2004.
[3] A. J. McKnight, A. S. McKnight, "Young novice drivers: careless or clueless?," Accident Analysis & Prevention, vol. 35 no. 6, pp. 921-925, DOI: 10.1016/S0001-4575(02)00100-8, 2003.
[4] M. A. Wetton, M. S. Horswill, C. Hatherly, J. M. Wood, N. A. Pachana, K. J. Anstey, "The development and validation of two complementary measures of drivers' hazard perception ability," Accident Analysis & Prevention, vol. 42 no. 4, pp. 1232-1239, DOI: 10.1016/j.aap.2010.01.017, 2010.
[5] L. Sun, R. S. Chang, "Hazard perception in young drivers: the role of hazard types," Journal of Psychological Science, vol. 40 no. 2, pp. 447-452, 2018.
[6] E. Constantinou, G. Panayiotou, N. Konstantinou, A. Loutsiou-Ladd, A. Kapardis, "Risky and aggressive driving in young adults: personality matters," Accident Analysis & Prevention, vol. 43 no. 4, pp. 1323-1331, DOI: 10.1016/j.aap.2011.02.002, 2011.
[7] J. Iskander, S. Hanoun, I. Hettiarachchi, M. Hossny, A. Bhatti, "Eye behaviour as a hazard perception measure," ,DOI: 10.1109/syscon.2018.8369509, .
[8] A. K. Mackenzie, J. M. Harris, "Eye movements and hazard perception in active and passive driving," Visual Cognition, vol. 23 no. 6, pp. 736-757, DOI: 10.1080/13506285.2015.1079583, 2015.
[9] P. J. Hills, C. Thompson, J. M. Pake, "Detrimental effects of carryover of eye movement behaviour on hazard perception accuracy: effects of driver experience, difficulty of task, and hazardousness of road," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 58, pp. 906-916, DOI: 10.1016/j.trf.2018.07.014, 2018.
[10] C. T. Scialfa, D. Borkenhagen, J. Lyon, M. Deschênes, "A comparison of static and dynamic hazard perception tests," Accident Analysis & Prevention, vol. 51, pp. 268-273, DOI: 10.1016/j.aap.2012.12.006, 2013.
[11] M. Sivak, "The information that drivers use: is it indeed 90% visual?," Perception, vol. 25 no. 9, pp. 1081-1089, DOI: 10.1068/p251081, 1996.
[12] P. R. Chapman, G. Underwood, "Visual search of driving situations: danger and experience," Perception, vol. 27 no. 8, pp. 951-964, DOI: 10.1068/p270951, 1998.
[13] D. Crundall, P. Chapman, S. Trawley, L. Collins, E. van Loon, B. Andrews, G. Underwood, "Some hazards are more attractive than others: drivers of varying experience respond differently to different types of hazard," Accident Analysis & Prevention, vol. 45, pp. 600-609, DOI: 10.1016/j.aap.2011.09.049, 2012.
[14] A. Borowsky, D. Shinar, T. Oron-Gilad, "Age, skill, and hazard perception in driving," Accident Analysis & Prevention, vol. 42 no. 4, pp. 1240-1249, DOI: 10.1016/j.aap.2010.02.001, 2010.
[15] D. E. Crundall, G. Underwood, "Effects of experience and processing demands on visual information acquisition in drivers," Ergonomics, vol. 41 no. 4, pp. 448-458, DOI: 10.1080/001401398186937, 1998.
[16] T. Falkmer, N. P. Gregersen, "Fixation patterns of learner drivers with and without cerebral palsy (CP) when driving in real traffic environments," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 4 no. 3, pp. 171-185, DOI: 10.1016/S1369-8478(01)00021-3, 2001.
[17] P. Konstantopoulos, D. Crundall, "The Driver Prioritisation Questionnaire: e," Accident Analysis & Prevention, vol. 40 no. 6, pp. 1925-1936, DOI: 10.1016/j.aap.2008.08.009, 2008.
[18] A. K. Pradhan, K. R. Hammel, R. DeRamus, A. Pollatsek, D. A. Noyce, D. L. Fisher, "Using eye movements to evaluate effects of driver age on risk perception in a driving simulator," Human Factors: The Journal of the Human Factors and Ergonomics Society, vol. 47 no. 4, pp. 840-852, DOI: 10.1518/001872005775570961, 2005.
[19] G. Underwood, P. Chapman, K. Bowden, D. Crundall, "Visual search while driving: skill and awareness during inspection of the scene," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 5 no. 2, pp. 87-97, DOI: 10.1016/S1369-8478(02)00008-6, 2002.
[20] G. Underwood, P. Chapman, N. Brocklehurst, J. Underwood, D. Crundall, "Visual attention while driving: sequences of eye fixations made by experienced and novice drivers," Ergonomics, vol. 46 no. 6, pp. 629-646, DOI: 10.1080/0014013031000090116, 2003.
[21] T. Dukic, T. Broberg, "Older drivers' visual search behaviour at intersections," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 15 no. 4, pp. 462-470, DOI: 10.1016/j.trf.2011.10.001, 2012.
[22] M. R. E. Romoser, A. Pollatsek, D. L. Fisher, C. C. Williams, "Comparing the glance patterns of older versus younger experienced drivers: scanning for hazards while approaching and entering the intersection," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 16, pp. 104-116, DOI: 10.1016/j.trf.2012.08.004, 2013.
[23] J. Werneke, M. Vollrath, "What does the driver look at? The influence of intersection characteristics on attention allocation and driving behavior," Accident Analysis & Prevention, vol. 45, pp. 610-619, DOI: 10.1016/j.aap.2011.09.048, 2012.
[24] M. Itoh, K. Yoshimura, T. Inagaki, "Inference of large truck driver's intent to change lanes to pass a lead vehicle via analyses of driver's eye glance behavior in the real world," pp. 2385-2389, DOI: 10.1109/SICE.2007.4421387, .
[25] C. Moran, J. M. Bennett, P. Prabhakharan, "Road user hazard perception tests: a systematic review of current methodologies," Accident Analysis & Prevention, vol. 129, pp. 309-333, DOI: 10.1016/j.aap.2019.05.021, 2019.
[26] D. Crundall, "Hazard prediction discriminates between novice and experienced drivers," Accident Analysis & Prevention, vol. 86, pp. 47-58, DOI: 10.1016/j.aap.2015.10.006, 2016.
[27] A. Gugliotta, P. Ventsislavova, P. Garcia-Fernandez, E. Peña-Suarez, E. Eisman, D. Crundall, C. Castro, "Are situation awareness and decision-making in driving totally conscious processes? Results of a hazard prediction task," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 44, pp. 168-179, DOI: 10.1016/j.trf.2016.11.005, 2017.
[28] K. A. Johnston, C. T. Scialfa, "Hazard perception in emergency medical service responders," Accident Analysis & Prevention, vol. 95, pp. 91-96, DOI: 10.1016/j.aap.2016.06.021, 2016.
[29] S. Malone, R. Brünken, "The role of ecological validity in hazard perception assessment," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 40, pp. 91-103, DOI: 10.1016/j.trf.2016.04.008, 2016.
[30] W. P. Vlakveld, "A comparative study of two desktop hazard perception tasks suitable for mass testing in which scores are not based on response latencies," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 22, pp. 218-231, DOI: 10.1016/j.trf.2013.12.013, 2014.
[31] J. Feng, H. Choi, F. I. M. Craik, B. Levine, S. Moreno, G. Naglie, M. Zhu, "Adaptive response criteria in road hazard detection among older drivers," Traffic Injury Prevention, vol. 19 no. 2, pp. 141-146, DOI: 10.1080/15389588.2017.1373190, 2018.
[32] L. Huestegge, E.-M. Skottke, S. Anders, J. Müsseler, G. Debus, "The development of hazard perception: dissociation of visual orientation and hazard processing," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 13 no. 1,DOI: 10.1016/j.trf.2009.09.005, 2010.
[33] C. T. Scialfa, D. Borkenhagen, J. Lyon, M. Deschênes, M. Horswill, M. Wetton, "The effects of driving experience on responses to a static hazard perception test," Accident Analysis & Prevention, vol. 45, pp. 547-553, DOI: 10.1016/j.aap.2011.09.005, 2012.
[34] A. Meir, T. Oron-Gilad, Y. Parmet, "Are child-pedestrians able to identify hazardous traffic situations? Measuring their abilities in a virtual reality environment," Safety Science, vol. 80, pp. 33-40, DOI: 10.1016/j.ssci.2015.07.007, 2015.
[35] J. S. Yeung, Y. D. Wong, "Effects of driver age and experience in abrupt-onset hazards," Accident Analysis & Prevention, vol. 78, pp. 110-117, DOI: 10.1016/j.aap.2015.02.024, 2015.
[36] P. Ventsislavova, D. Crundall, T. Baguley, C. Castro, A. Gugliotta, P. Garcia-Fernandez, W. Zhang, Y. Ba, Q. Li, "A comparison of hazard perception and hazard prediction tests across China, Spain and the UK," Accident Analysis & Prevention, vol. 122, pp. 268-286, DOI: 10.1016/j.aap.2018.10.010, 2019.
[37] M. A. Wetton, A. Hill, M. S. Horswill, "The development and validation of a hazard perception test for use in driver licensing," Accident Analysis & Prevention, vol. 43 no. 5, pp. 1759-1770, DOI: 10.1016/j.aap.2011.04.007, 2011.
[38] N. Kinnear, S. W. Kelly, S. Stradling, J. Thomson, "Understanding how drivers learn to anticipate risk on the road: a laboratory experiment of affective anticipation of road hazards," Accident Analysis & Prevention, vol. 50, pp. 1025-1033, DOI: 10.1016/j.aap.2012.08.008, 2013.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Copyright © 2021 Jingshuai Yang et al. This work is licensed under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Aiming at young drivers’ hazard perception (HP) and eye movement, a cross-sectional study was conducted in the city of Xi’an, China. 46 participants were recruited, and 35 traffic scenes were used to test drivers’ hazard perception and eye movement. The difference analysis and correlation analysis were carried out for the acquired data. The results suggest that some indices of hazard perception and eye movement are significantly correlated. A higher saccade speed is in the direction of higher hazardous scenes. Higher complex scenes result in smaller saccade angle. The number of hazards unidentified is negatively influenced by complexity degree and hazardous degree of traffic scenes, and similar associations are found between hazard identification time, complexity degree, and hazardous degree. The hazard identification time and the number of hazards slowly identified are positively affected by the number of fixations and the number of saccades. Meanwhile, differences in the hazardous degree evaluation, hazard identification time, number of hazards unidentified, number of fixations, and number of saccades are found in different types of traffic scenes. The results help us to improve the design of road and vehicle devices, as well as the assessment and enhancement of young drivers’ hazard perception skills.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details

1 School of Automobile, Chang’an University, Xi’an 710064, China
2 School of Automobile, Chang’an University, Xi’an 710064, China; The Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, Shanghai 201804, China