Keywords

1 Introduction

1.1 The Concept of Situational Awareness

The concept of situation awareness was first proposed and developed in the aviation field. Although SA has some similarities in the area of flight, driving definitions of SA are still needed in the area of driving to identify potentially influential Mission, environment and personal factors [1]. Situation awareness is the operator’s perception of the elements in the environment within a given time and space, his or her understanding of the meaning, and the prediction of its future state of development [2].

In traffic psychology, situation awareness is the driver’s perception, understanding, and prediction of many elements of the vehicle’s condition, traffic signs and signals, traffic information, weather conditions, etc. throughout the road environment. In general, driving task involves five-phased of information processing function, including perception, comprehension and projection, as well as a decision on a course of action and implementing the action. The perception, comprehension and projection functions are the basis for driver situation awareness [3]. In traffic psychology, situational awareness refers to drivers’ perception, understanding and prediction of vehicle status, traffic signs and signals, road information, weather conditions and other elements in the whole road environment. At present, the research on driver’s situational awareness is roughly divided into three points of view. The first is from a psychological point of view, it focuses on the study of situational awareness as a simple internal cognitive phenomenon, which is mainly related to the research and analysis of drivers. The second is from an engineering point of view, it focuses on the study of situational awareness through different vehicle technology and road infrastructure. It mainly analyzes artifacts. And the third is from a systematic engineering point of view, it focuses on the study of situational awareness from the interaction of drivers, artifacts, and the interaction between both of them [4].

Three-level model of situation awareness is first proposed by Endsley. Endsley divides situational awareness into three different levels of processing (Fig. 1) [2].

Fig. 1.
figure 1

Situational awareness processing

Full size image

In the first level of situation awareness, the driver perceives the various factors in the surrounding environment (perception). In the second level, the driver needs to combine his past relevant experience with the key elements of the surrounding environment, such as traffic environment and flight environment, to form a kind of coherent picture to understand (comprehend). In the third level, drivers make their own predictions about the future development and trends of these information elements (projection). Endsley’s situational awareness model is a typical feedback loop model in the field of situational awareness. It is the situational awareness that feeds back the current state of the environment to the individual. Then, the situational awareness feedbacks the result of integrating the current environment to the individual decision-making mechanism and makes an appropriate behavioral response. When the behavioral response can change the current unfavorable environmental state. The entire feedback loop is over.

The research by Ma investigated the effects of an adaptive cruise control (ACC) system, and cell phone use in driving, on a direct and objective measure of SA, and assessed the competition of multiple driving and communication tasks for limited mental resources in terms of driving performance. Results indicated use of the ACC system to improve driving task SA under typical driving conditions, and to reduce driver mental workload. However, the cell phone conversation caused deleterious effects on driving SA and increased driver mental load [3].

Matthews et al. proposed a contextual awareness information processing model based on the three-level model of contextual awareness. According to the different mission objectives they divided the driving behavior into three categories. The first is operational driving, its meaning is to ensure that the normal operation of their vehicles by the operation. For example, driving a car on the road to pay attention to traffic lights and speed limit signs. The second is tactical driving, it refers to what the driver does to ensure safe interaction with other vehicles. Such as driving cars on the road to overtake and change lanes. The third category is strategic driving, it refers to the driver of the known information of high-level reasoning, calculation process of thinking about driving strategy. For example, acceleration is slowed down in advance. We can see that operational driving relies on the first level of the three-level model, tactical driving relies on the second and third levels, and strategic driving relies on all three levels [5].

Salmon et al. found that A high level of situational awareness does not necessarily depend on a low level of situational awareness. Some very experienced drivers can skip a low level of situational awareness directly to get a higher level of situational awareness [6].

Walker et al. found that an approach to SA based on Neisser’s perceptual cycle theory is anchored to a network based methodology. It is applied within the context of a longitudinal on-road study involving three groups of 25 drivers, all of whom were measured pre- and post-intervention. One experimental group was subject to advanced driver training and two further groups provided control for time and for being accompanied whilst driving. Empirical support is found for all five hypotheses. Advanced driving does improve driver SA but not necessarily in the way that existing situation focused, closed loop models of the concept might predict [7].

Matthews et al. outlined multiple elements of awareness defining SA in driving, including spatial awareness, identity awareness, temporal awareness, goal awareness and system awareness. They said spatial awareness refers to an appreciation of the location of all relevant features of the environment. Identity awareness refers to the knowledge of salient items in the driving environment. Temporal awareness refers to knowledge of the changing spatial “picture” over time. Goal awareness refers to the driver’s intention of navigation to the destination, and the maintenance of speed and direction. System awareness refers to relevant information on the vehicle within the driving environment, which may also be viewed as a system [5].

Gugerty and Tirre presented a similar concept of driver situation awareness. They said drivers must maintain navigation knowledge, local scene comprehension (knowledge of nearby traffic for maneuvering), knowledge of spatial orientation, and knowledge of their vehicle’s status to maintain good SA during driving [8].

Gugerty and Tirre and Matthews et al. considered in-vehicle system interaction knowledge to be important in a driving environment, for example, when a car traveling at a constant speed under cruise control enters a higher speed limit area, driver awareness of their vehicle speed, the speed limit and knowledge of how to set a higher speed represents good SA [9].

Crundall and Underwood found that the differences between novices and experienced drivers in their distribution of visual attention under different levels of cognitive load imposed by different types of road, and as reflected in their visual search strategies. The results suggested that experienced drivers select visual strategies according to the complexity of the roadway, and that the strategies of novices are too inflexible to meet changing demands [10].

1.2 The Main Method of Measuring Situational Awareness

The current main method of measuring situational awareness is divided into situation awareness global assessment technique, eye movements, and Proposition networks. It features a situational awareness-inducing in a driving simulator. In the driving simulator, testers freeze the simulated tasks of the subjects at random time points, and present blank screens, allowing subjects to answer some questions referred to the task according to their memory or to make a detailed description of a specific situation. eye movements reflect the driver’s traffic information for different degrees of processing through the eye movement indicators. This is the most commonly used measurement method for studying driver situational awareness. Eye movement measurement refine the driver’s cognitive process, to more accurately examine the difference between different drivers situational awareness, such as experienced drivers and novices between the level of awareness of situational awareness and support for driver’s situational awareness training. Proposition networks refers to use specialized software to analysis and process verbal reports of drivers and cognitive tasks after the interview, and then form a network model that contains a variety of informational elements. We describe the driver’s situational awareness through the development and changes of this network model. In Proposition networks, We use the ellipse to represent the information elements in the environment, and the labeled arrows indicate the relationship between the elements [11].

2 Method

Through indoor simulated driving, DRT, timer and subjective evaluation table are used to collect data. The static experimental scene uses the PC-side driving simulation software to simulate the driving scene. It is divided into Fig. 2 (single road, multiple straights, short distance) and Fig. 3 (complicated road, multiple curves, long distance).

Fig. 2.
figure 2

Single road, multiple, straights, short distance

Full size image
Fig. 3.
figure 3

Complicated road, multiple curves, long distance

Full size image

11 subjects were selected, all 21–23 year-old college students, have C1 driver’s license and can skillfully drive a car. First of all, we confirm their personal information, driving experience and driving habits, to ensure that the subjects were healthy and no discernible hearing disorders. Before the experiment, the training of all subjects were consistent, so that subjects were familiar with the experimental background in the same grade. The training process is to first explain to the subjects experimental principle, experimental process, experimental purpose, experimental interface, operation method. Then subjects were simulated driving, familiar with the driving environment, to meet the experimental requirements (see Fig. 4).

Fig. 4.
figure 4

The training process

Full size image

In addition, two helpers were required for each experiment, one person was in charge of control and data recording of the DRT measurement equipment, and the other was responsible for voice prompts the subjects during the experiment. The experiment takes two experimental variables: different task complexity and different information push on the situational awareness of drivers.

Information push mode is divided into three types: First of all in a single unfamiliar environment, the subjects observed the field work independently and completed the task, without the information hints of the task. Second subjects complete the field task with the aid of the task path planning map. Third subjects complete the task under the real-time path voice prompt.

This experiment is a two-level, two-level and a three-level (2 * 3) mixed experiment, a total of six experiments, as shown in the Table 1.

Table 1. Experimental design table
Full size table

We numbered the experiment in order to facilitate the recording. The first digit of the experiment number represents the level of task complexity, 1 for low level and 2 for high level. The second digit represents how information is pushed, 0 for traffic observation, 1 for the map tips, 2 for voice prompts. In order to prevent the experimental order may give the subjects the learning effect, we require different subjects have different experimental sequences. In other words, we randomized the experimental sequences, and the same scene will not be tested twice in succession.

The dependent variables in the experiment include subjective evaluation of situational awareness, DRT performance and driving performance. During the experiment, the experiment process was recorded by the camera to analyze after the experiment.

2.1 Subjective Evaluation and Analysis of Situational Awareness

After the experiment, we explained the meaning of situational awareness to the subjects. When the subjects got a good understanding of the subject, we asked the subjects to score the scores according to Table 2.

Table 2. Subjective rating
Full size table

2.2 DRT Performance

Obtain the driver’s reaction ability that the reaction time and the target hit error rate through the DRT performance.

2.3 Driving Performance

Get the driving performance through the driver to complete the task of driving the situation, driving performance is divided into driving task completion time and driving the number of collisions.

3 Results and Discussion

The results obtained by the above experiment, the analysis method used is EXCEL scatter plot and histogram roughly analyzed, and then we judge the significance of the experimental results by the hypothesis t-test.

3.1 Subjective Evaluation and Analysis of Situational Awareness

After the experiment, we explained the meaning of situational awareness to the subjects. When the subjects got a good understanding of the subject, we asked the subjects to score the scores according to Fig. 5.

Fig. 5.
figure 5

Subjective evaluation

Full size image

We averaged the subjective evaluation scores for each type of task, as we can see from the results, when no prompt was provided and the subjects only relied on the road to obtain information, the subjects rated the scores as 4.7 and 5.8, they were neutral to difficult. When there were map prompts, the scores dropped to 3.4 and 4.5, which was slightly easier than without information push, but not obvious. However, when there was a voice prompt, the evaluation score reached 2.4 and 3.7, significantly reduced. It proves that when the other conditions are the same, the voice prompts can make the subjects better perceive the traffic information. Because subjects are required to divert part of the visual resources to get information from the map prompts and need to locate the vehicle after obtaining the information to determine their location before making decisions. When they use the voice prompts, visual resources will not be occupied, just get the voice information, compared with the current environment, they can make the appropriate decisions. Without any information push, the subjects need to observe the road conditions, pick out the information they need from a lot of information, and then perceive, understand and predict, so in this case, the subject’s situation awareness rating will be high, this indicate a high level of situational awareness will make the driver’s subjective evaluation score lower. By t test, all P << 0.05. Therefore, the judgment of subjective evaluation is very significant, it has statistical significance (Tables 3, 4, 5 and 6).

Table 3. Subjective evaluation t-test mean
Full size table
Table 4. Subjective evaluation t-test mean
Full size table
Table 5. Subjective evaluation t-test mean
Full size table
Table 6. Subjective evaluation t-test mean
Full size table

3.2 DRT Performance Analysis

The DRT reaction time is the time difference between the moment when the driver presses the reaction button and the diode starts to emit light after the driver sees the light-emitting stimulus. Response time under different message prompts, all of which are average response times. We use the DRT reaction time as the vertical axis, subjective evaluation of situational awareness as abscissa, made of a scatter plot (Figs. 6 and 7). All of response time are average response time.

Fig. 6.
figure 6

Subjective evaluation - DRT scatter plot in scene 1

Full size image
Fig. 7.
figure 7

Subjective evaluation - DRT scatter plot in scene 2

Full size image

From the figure we can see that with the increase of situational awareness scores, the response time of subjects showed a downward trend. The results showed that in the driving process, when prompted, people’s response time will be longer. Without information push, the response time of the subjects in both scenarios was 534.29 ms and 610.26 ms. The reaction time at the map prompt is 596.11 ms and 665.83 ms and the response time at voice prompts was 667.31 ms and 717.86 ms. The reason for the longer reaction time is that on the one hand, map prompts and voice prompts can distract users thus prolonging the reaction time, on the other hand, observing the scatterplot of situational awareness and response time, we can clearly see that with the increase of situational awareness scores, subjects’ response time is reduced. Because in a strange scene, if the subject is too difficult to obtain situational information, which level of situational awareness is not enough, he can not be better integrated into his driving work. So he has enough energy on the DRT mission. As the level of situational awareness improved, subjects were able to access information easily, so he could dedicate himself to driving, his energy spent on DRT tasks became less and his reaction time lengthened. This shows that a high level of situational awareness leads to longer driver DRT response time.

3.3 Driving Completed Performance Analysis

During driving, the driver is not aware of the surrounding information (lack of situational awareness) or is wrongly aware of the surrounding information (false situational awareness), resulting in the occurrence of a collision. We put each crash as a mistake, for performance evaluation (see Fig. 8).

Fig. 8.
figure 8

Frequency

Full size image

The results show that information prompts will make the driver effectively avoid collision during driving. Performance is 4.72 times and 10.82 times without any information push. 2.91 times and 4.91 times for map prompts, 2.55 times and 5.36 times for voice prompts. So under the information push, subjects were able to obtain timely and accurate information on the road, make timely adjustments to reduce the collision. By observing the video, the number of collisions that occurred at the corners under voice prompts was more than the number of prompts at the map. Because the subjects passed the curve, they were prompted by the map to have a clear understanding of the curve. However, they can only get the message that the next corner will be through voice prompts, and there is no intuitive understanding of the corner in the brain. In Scene 2, there are many corners, so the performance of the map prompts is better than the voice prompts. At the same time, we found through observation that in sharp turns, the driving performance of the voice prompt is better than the driving performance of the map. This is due to that voice prompts can convey information to the driver promptly and clearly before turning, but map information acquisition depends on whether the driver has checked the map at the turning point. Therefore, for the continuous complex road conditions, the map prompts can optimize the driver’s situational awareness. For some sudden and other unexpected traffic conditions, under the voice prompts, the driver’s situational awareness is better. For the continuous complex road conditions, the driver can be prompted by voice to observe the map, and at the same time the map prompts, so as to enhance the driver’s situational awareness.

4 Conclusion

The results show that under the same conditions, the level of awareness of the driver under the prompt of the voice is higher than that of the driver under the forerunner of the mission path planning, and the level of situational awareness under the simple operation environment is higher than that of the complex operation environment. By analyzing the performance of work, we find that with the improvement of the level of situational awareness, the time of the task is shorter and the performance of the work becomes better.

This research is of great significance to training and improving the level of awareness of drivers.