Abstract
Objective
This study investigated how the visualization of an ecological interface affects its subjective and objective usefulness. Therefore, we compared a simple 2D visualization against a contact-analog 3D visualization.
Background
Recently, head-up displays (HUDs) have become contact-analog and visualizations have been enabled to be merged with the real environment. In this regard, ecological interface design visualizing boundaries of acceptable performance might be a perfect match. Because the real-world environment already provides such boundaries (e.g., lane markings), the interface might directly use them. However, visual illusions and undesired interference with the environment might influence the overall usability.
Method
To allow for a comparison, 49 participants tested the same ecological interface in two configurations, contact-analog (3D) and two dimensional (2D). Both visualizations were shown in the car’s head-up display (HUD).
Results
The driving simulator experiment reveals that 3D was rated as more demanding and more disturbing, but also more innovative and appealing. However, regarding driving performance, the 3D representation decreased the accuracy of speed control by 6% while significantly increasing lane stability by 20%.
Conclusion
We conclude that, if we want environmental boundaries guiding our behavior, the indicator for the behavior should be visualized contact-analog. If we desire artificial boundaries (e.g., speed limits) to guide behavior, the behavioral indicator should be visualized in 2D. This is less prone to optical illusions and allows for a more precise control of behavior.
Application
These findings provide guidance to human factors engineers, how contact-analog visualizations might be used optimally.
Introduction
Since the late 20th century, head-up displays (HUDs) have become increasingly popular in the context of automotive research and development (Gish & Staplin, 1995; Kiefer, 1998, 2000). Their usage was pioneered for fighter jets. However, today, HUDs are becoming increasingly available for end users in vehicles of all kinds. Further, HUDs have recently become contact-analog and the information displayed in the screen has been enabled to be merged with the real environment in a way that the HUD content is spatially related to the outside world. Therefore, the displayed artificial objects refer to objects in the real-world background. For the driver, a fusion is created between the real-world and virtual objects in the HUD.
Considerable literature suggests using such a contact-analog HUD for diverse objectives, such as navigation purposes or hazard warnings (Park & Park, 2019). Moreover, ecological interface design (EID) is a perfect match. Ecological interfaces aim to visualize the boundaries of acceptable performance to allow direct perception-action mappings and thus behavior based on the lowest cognitive control level possible (Burns & Hajdukiewicz, 2004; Vicente, 2002; Vicente & Rasmussen, 1992). In contrast, a conventional interface would present information in an alphanumeric format which then demands more cognitive control. Because the real-world environment already provides such boundaries of acceptable performance (e.g., lane markings or preceding cars) an ecological interface might use them and align its contact-analog presentation. Schewe and Vollrath (2019) designed and evaluated an ecological interface for speed control and named it distance speedometer. As shown in Figure 1, the driving speed is visualized by the length of a blue bar and the control speed is visualized by a white bracket. The upper and lower end of the bracket represent a range of favorable speeds (±5 km/h), while the middle represents the preferred speed at a certain time (e.g., 50, 70, or 100 km/h).
Ecological distance speedometer (top) in comparison to a conventional speedometer (bottom).
Schewe and Vollrath (2020) compared its effect against a conventional digital speedometer in the HUD and demonstrated that a significant reduction in cognitive load was achieved while enhancing the driving performance. In contrast to the conventional interface, Schewe and Vollrath (2020) claim that the ecological representation increased the accuracy of speed control by 6% while significantly increasing lane stability by 15%. However, they did not investigate whether the ecological interface for speed control should be augmented or whether the ecological design itself is superior to the conventional interface. Therefore, the question remains regarding whether the ecological interface should be augmented or whether a contact-analog visualization tradeoff exists due to human perception. This question was the objective of the paper at hand.
Technical problems with contact-analog visualizations are well known. Road surfaces and the car’s position always have to be matched and the digital overlay might lose its connection to the real world (Sadovitch, 2020). Visual illusions and other problems with human perception of space might also be major issues. As Gibson and Walk (1960) demonstrated in their experiment “the visual cliff,” perception of depth can be seen as inherent. Babies were to crawl to their mother over a glass plate with a visual cliff, but they did not dare cross over. Therefore, since depth perception seems innate, the factor of depth might theoretically not influence the cognitive demand posed by the augmented interface. That means that the demands on working memory capacity generated by the information being examined might not change if the same interface is visualized 3D contact-analog instead of two dimensional.
However, regarding perception and the relation to a possible loss in performance, plenty of known inaccuracies are associated with monocular depth cues and motion-induced depth cues. As summarized by Teittinen (1993), for example, image-induced masking and overlapping result in the perception of something located farther away. Moreover, parallel lines are perceived as converging lines in the distance, which is another monocular cue for depth. The relative size in the field of vision is also important because objects that physically have the same shape are estimated to be farther away if they take up less space in the field of vision. Therefore, smaller objects, such as motorcycles, are perceived as farther away than larger objects. The closer an object is to the horizon line, the farther away it appears to be. Generally, objects represented farther up in the field of view (i.e., at a higher position) are seen as being farther away. Moreover, the texture gradient seems more densely packed with increasing distance. Of course, the size of known objects also influences the estimation of the distance and the size perception of unknown objects (Epstein, 1965).
On the side of motion-induced depth stimuli, the motion parallax is the most important depth cue (Rogers & Graham, 1979). Nearby objects appear to move quickly, and distant objects seem to move slowly. Because most HUDs still offer no stereopsis, these cues are not discussed further. Moreover, stereopsis seems to only have a positive effect on driving performance in specific situations, which are highly dynamic (Bauer et al., 2001).
To summarize, the perception of depth might interfere with the augmentation and thus reduce its effective precision. Whether such interference is the case will be investigated in the study at hand. However, these effects could also help to guide behavior. Accordingly, it is worth researching a possible tradeoff and discussing whether ecological interface concepts should use a 3D contact-analog display, or whether a simple 2D display is as effective or even has some advantages due to better perceivability.
Therefore, this study tested the same ecological interface in two display configurations, contact-analog (3D) and 2D.
Specifically, the following research questions are examined:
Is the speed control worse with a 3D representation of the ecological interface (due to optical difficulties)?
Does the 3D representation improve lane stability? (Does it make sense to visualize a contact-analog reference to the street boundaries next to the main purpose of speed regulation?)
Is the workload equally low for 2D and 3D?
Is there a difference in acceptance?
As explained for the ecological interface design, the cognitive workload should be reduced when using the ecological interface in any way. However, this aspect must be quantified for a fair comparison. Moreover, in trying to answer the overall research questions, the driving performance should be analyzed. Lastly, qualitative analyses of both interfaces should be examined to assess individual opinions. The following measures were assessed for a sound comparison of both representations: Acceptance (usefulness and satisfaction), subjective mental effort, cognitive workload, and a qualitative content analysis of comments on both visualizations.
Method
Design and Procedure
The experiment was designed to allow for a comparison of the contact-analog (3D) and 2D display of the ecological human machine interface (HMI). Therefore, both scenarios were presented in a controlled order, meaning that half of the participants first saw the 2D scenario and half of the participants the 3D scenario, ensuring that effects of the order were canceled out. Before each scenario, participants received an instruction on how to use the respective interface and performed a familiarization run in which they had to accelerate and decelerate according to the visualizations. As shown in Appendices A and B, participants were instructed to keep the speed while driving as naturally as possible. After each driving scenario the subjective scales were queried. At the end of the whole experiment, the direct comparison of both interfaces was conducted.
Displays
As illustrated in Figure 2 and explained in appendices A and B, the same ecological interface design approach was used in two display configurations. Both displays are based on the same ecological interface design approach and, thus, function in a similar way.
Two dimensional representation vs. three dimensional representation of the ecological interface. The blue bar represents the current driving speed and the white bracket indicates the desired set speed.
The blue bar, at the end of a semitransparent carpet, represents the driving speed by its vertical position (either 2D in the HUD or 3D on the road). For the 3D position, the driving speed is multiplied by a time headway of 1 s. That means, the 3D display uses the current driving speed to predict where the car will be 1 s into the future and displays it on the road. In the few curves, the display’s transparency gradually was increased by 10%, because otherwise the bar, which is always straight ahead of the car, might have obstructed the view on the road markings. For the 2D display, the position of the blue bar does not forecast the position of the car. Rather the height of the bar corresponds to the speed of the car and is displayed on a vertical scale in front of the driver. However, since the 2D visualization is not augmented anymore, the lateral guidance, meaning the car’s position in contrast to the boundaries of the street, is not provided anymore.
The set speed, meaning the optimal driving speed, is visualized by the vertical position of the white bracket. Its middle position indicates the optimal driving speed (e.g., 50, 70, or 100 km/h) and the brackets represent ±5 km/h around the optimal speed. For 2D, the current speed limit corresponds to the height of the white bracket on the vertical scale. For 3D, the augmented position is determined in a similar way as for the blue bar. If the driving speed matches with the set speed, the blue bar and white bracket match perfectly, if the blue bar exceeds the bracket (which indicates a range of ±5 km/h around the optimal driving speed), the driving speed is too high.
Table 1 shows the speed related display sizes on the simulator’s screen. To reduce effects of image size, the mean size for both displays was designed to be similar. However, due to the different type of displays (2D and 3D), size differences at 50 and 100 km/h were inevitable and part of the experiment.
Image Sizes on the Simulator’s Screen in M
Scenario
The basic driving environment consisted of a rural road. Accordingly, speed limits of 50, 70, and 100 km/h were implemented. Three complete curves (Curves had a length of 190 m and a radius of 700 m; Length of all curves together equals 4% of the track) were integrated, but the speed changes only occurred in straight sections. Each ride took about 12 min and included a familiarization run. Figure 3 depicts the track, with its speed limits and curves.
The track with speed limits in kilometers per hour.
Moreover, no crosswinds and just one to two cars on the opposite side of the street every minute were simulated. Figure 4 depicts the dimensions of the simulated road.
Dimensions of the simulated road.
Participants
The sample includes 49 participants who were recruited from an internal university database. Six of the initial 55 participants had to be excluded due to technical failure where the simulation either stopped working or the data recording was faulty (five exclusions) or due to simulator sickness (one exclusion). Table 2 presents the participants’ demographics. The age range was biased toward younger drivers, as 42 participants were between 19 and 49 years old, while only 7 participants were between 50 and 85 years old.
Demographics of the Sample
Note. M = mean; SD = standard deviation.
The research complied with the German Psychological Societies’ adaptation of the American Psychological Association’s Code of Ethics and was approved by the Institutional Review Board of the TU Braunschweig. Informed consent was obtained from each participant.
Simulator and Driving Performance
The fixed-base simulator of the Department of Engineering and Traffic Psychology at TU Braunschweig consists of a physical mock-up cockpit and an array of three screens (2 × 2 m), in addition to the corresponding LCD projectors (1920 × 1080 pixels each) that cover a 180° field of vision from the driver’s seat (approximately 2 m away from the center of the front screen). The HUD was simulated. Figures 5 and 6 present the configuration as well as the running simulation.
Simulator setup.
Running simulation.
Two seven-inch screens were used as side mirrors. An automatic gearbox was simulated, and the steering wheel, gas pedal, and brake pedal were equipped with force feedback. Driving sounds were played by a sound system and SILAB 5.0 (WIVW GmbH, https://wivw.de/en/silab) was used as simulator software.
The lateral driving performance was assessed using the standard deviation of the lateral position (SDLP), which represents “weaving” around a middle trajectory (O’Hanlon et al., 1982). To measure the longitudinal driving performance, the standard deviation of the speed deviation from the speed limit was analyzed. If someone always drives five km/h too fast, but this in a very constant way, this reflects a high level of control anyway.
Cognitive Load
To assess the cognitive load, we used the detection-response task, described in ISO 17488 (International Organization for Standardization, 2016). It measures the attentive effects of cognitive workload induced by a secondary task (Krause et al., 2014). The detection-response task used for the paper at hand, provided a vibration next to the clavicles of the participants. It was implemented by a mobile application (https://github.com/ InstituteOfErgonomics/MDT) on a Samsung Galaxy S3 smartphone from 2013. A physical button was attached to the steering wheel and the participants were instructed to press as quickly and as accurately as possible to every vibration they detected. As described in the ISO, the interstimulus interval varied randomly from 3 to 5 s and the response signals remained on for 1000 ms or until the button was pressed. The reaction times and hits were used as relevant measures for the cognitive load.
Subjective Assessments
The subjective effort was measured using the single item 220 mm long (barely demanding–extremely demanding) visual-analog SEA scale (“Skala zur Erfassung subjektiv erlebter Anstrengung”; Eilers et al., 1986), a German translation of the Subjective Mental Effort Questionnaire (SMEQ; Zijlstra & Van Doorn, 1985).
To assess mental stress, the Driving Activity Load Index (DALI) was used (7 items; for example, how high were the visual demands?). The DALI is a survey instrument based on the NASA TLX to assess the mental stress of a driver in road traffic (Pauzié, 2008).
Acceptance was measured using the bipolar Van der Laan scale. It contains the subcategories of usefulness and satisfaction (Van Der Laan et al., 1997). The 9 items and their belonging to the subcategories are listed in Table 3.
Acceptance Scale From Van Der Laan, Heino and De Waard (1997)
Moreover, a direct comparison of 2D and 3D on a 15-point Likert-Type scale was used to assess the following seven aspects: helpful, disturbing, demanding, innovative, supporting, appealing, and distracting. Counted from left to right, a score of eight represents a neutral rating. Figure 7 shows an example of the questionnaire allowing the direct comparison of both visualizations. Participants first had to choose one of the five top categories and then refine their rating. The final score was derived by assigning numbers to the 15 subcategories.
Example for the questionnaire allowing direct comparisons.
Lastly, the participants were asked to comment on the comparison of the 3D representation versus the 2D representation in free text.
Statistical Analysis
The data recorded for both display scenarios were analyzed using a one-way repeated-measure multivariate analysis of variance (MANOVA) for each construct. The data from the direct comparisons were analyzed using one sample t-tests (e.g., Gerald, 2018).
Qualitative Analysis
The qualitative analysis followed the process described by Mayring (1994). In the beginning, the text material was read to obtain a first overview. Then, the participants comments were divided into comments focusing on 2D and comments on 3D. Next, as Mayring suggests, all unnecessary, repetitive, or affirmative words were deleted (e.g., “I think,” and “rather”). This was followed by a further division of the comments into statements. Based on these statements, meaningful categories were first formed and then concretized based on keywords. Then, all individual statements were assigned to one of these categories. Finally, the frequencies of positive and negative evaluations were compared. Two independent raters confirmed the analysis.
Results
The results are structured into two parts. Part one contains the quantitative analysis of the driving data. The second part consists of the analysis of the qualitative data, including the content analysis according to Mayring.
Driving Performance
For the driving data, we analyzed whether a statistically significant difference exists in driving performance, which is measured in terms of SDLP and the SD of speed deviations, caused by the displays. A statistically significant difference was found between the displays on the combined dependent variables of driving performance, F(2, 47) =18.99, p < .01; Wilks’ Λ = .553; partial η2 = .447. A follow-up univariate t-test for repeated measures indicated that the SDLP was statistically different between the conditions of display, t(48) = 31.76, p < .01; partial η2 = .40. For the SD of speed deviations, another difference could be found, t(48) = 6.36, p = .02; partial η2 = .12, however with a smaller effect size. Figure 8 depicts the differences concerning the SDLP and SD of speed deviations. As can be seen, there is a more precise lane keeping with 3D (2D: M = .15, SD = .04; 3D: M = .12, SD = .03) and a more precise speed keeping with 2D (2D: M = 8.03, SD = 1.26; 3D: M = 8.53 SD =1.66).
Standard deviation of the lateral position (SDLP) and standard deviation of speed deviations (*p < .05; mean and SE).
Cognitive Load
To assess the displays influence on cognitive load, we analyzed whether a statistically significant difference in cognitive load, measured in terms of reaction times in ms (2D: M = 622, SD = 189; 3D: M = 628, SD = 189) and hit rates in % (2D: M = 86, SD = 13; 3D: M = 86, SD = 14), is caused by the factor of display. No statistically significant difference between the displays influence on the combined dependent variables representing cognitive load was found, F(2, 47) =.12, p = .90; Wilks’ Λ = .995; partial η2 = .005.
Subjective Assessments
The SEA scale was analyzed using a t-test for repeated measures, which indicated no statistically significant difference between the display s on the perceived effort, t(48) = 2.74, p = .11; partial η2 = .05. Both ratings indicate a medium amount of subjective effort to use the displays (2D: M = 108, SD = 52; 3D: M = 99, SD = 53).
Moreover, no statistically significant difference was found between the displays on the dependent variables for the DALI (effort of attention, visual demand, auditory demand, temporal demand, interference, situational stress), F(8, 41) =.86, p = .55; Wilks’ Λ = .85; partial η2 = .15, indicating no substantial differences in mental stress (2D: M = −3.25, SD = 1.12; 3D: M = −3.23, SD = 1.13).
Analyzing the Van der Laan scale, no significant difference was found between the displays on the combined dependent variables for the system acceptance, F(2, 47) =2.69, p = .08; Wilks’ Λ = .897; partial η2 = .10. However, follow-up t-tests for repeated measures indicated that satisfaction (2D: M = .06, SD = .73; 3D: M = .28, SD = .77) was statistically different between the conditions of display, t(48) = 5.03, p = .03; partial η2 = .10. For usefulness (2D: M = −.64, SD = .69; 3D: M = −.44, SD = .71), t(48) = 3.28, p = .08; partial η2 = .06, no significant differences could be shown. As Figure 9 reveals, the ratings are in the middle range of acceptance.
Usefulness and satisfaction (*p < .05; mean and SE).
Analyzing the direct comparison of both visualizations, one sample t-tests against the neutral rating of 8 were performed. As the following Table 4 indicates, four significant deviations were found. A value of 1 indicates that 2D is most likely to represent a certain attribute, and in contrast, a value of 15 indicates that 3D represents this attribute. In accordance, 3D is perceived as being more disturbing, demanding, and distracting. However, in a direct comparison, 3D is also rated as more innovative.
Direct Comparison
Note. *Significant.
Qualitative Content Analysis
When reading the following sections, please note the relatively small number of statements. Therefore, we advise comparing the frequencies rather than the percentages. Dividing the 49 participants’ comments into separate statements, out of 103 statements, 63 (61.2%) refer to the 3D, and 40 (38.8%) refer to the 2D representation. Of these, 58 (56.3%) are negative, and 45 (43.7%) are positive. Viewed individually, 43 of 63 statements relating to 3D are negative (68.3%) and 20 (31.7%) are positive. For 2D, a total of 15 (37.5%) out of 40 statements are negative, and 25 (62.5%) are positive. A χ2 test of independence was performed to examine the relation between the displays and the qualitative feedback. The relation between these variables was significant, X2 (1, N = 103) =9.41, p < .01. Overall, 3D is less likely than 2D to be rated positive. Finally, the following categories were identified: Attention, Presentation, Control and unspecific statements. In the following sections, the differences in the evaluations of the two displays are described category by category, as also depicted in Figure 10.
Qualitative analysis of both displays. Count of positive and negative statements for the categories of attention, presentation and control.
Attention
All statements that have been assigned to this category refer in some way to attention, concentration, effort, and distraction of the driver from the main driving task (keeping speed, etc.), caused by the displays. Keywords include strenuous, concentrate, much attention, and distracted. There were 23 (22.3%) statements that could be assigned to this category. Of these, 16 (69.6%) are negative, and 7 (30.4%) are positive. Concerning the statements regarding 3D, a distribution of 3 (21.4%) positive statements to 11 (78.6%) negative statements was observed. In 2D, the ratio is more balanced (ratio of 4 [44.4%] positive to 5 [55.5%] negative).
The statements indicate that the interface in 3D is more distracting because it draws more attention to the road (or display) than to the traffic. The display (bar) goes deeper and takes up more space, which may focus attention on it. The 2D display seems to be easier to follow. It requires less attention and less focus.
Presentation of the Interface
This category refers to the drivers’ comments about the interface in general: Whether it was well designed, in the right place, or the right size, or how it was displayed in the curves. The keywords for this were “covering,” “transparent,” “too large/small,” “pleasant,” “display/bar.” A total of 39 (37.9%) statements were made regarding the display; 19 (48.7%) are negative and 20 (51.3%) are positive. Examining the different displays, twice as many statements were made about the 3D display than the 2D display (3D: 26 [66.6%]; 2D: 13 [33.3%]). Concerning the statements on 3D, a total of 15 (57.7%) are negative and 11 (42.3%) are positive. The most frequent negative judgment was that the 3D interface was transparent in the curves and that this affected the driving experience. In addition, it covered much of the road, which was also perceived negatively. However, the positive statements emphasize that the 3D display was better integrated (except for the curves) and therefore was more in the focus. In the 2D condition, the ratio is clearer: 4 of 13 (30.8%) statements are negative while 9 (69.2%) are positive. The statements said that the 2D display was not well integrated and distracted them but that it was considered simpler. Despite the positive assessment of the simplicity and unambiguousness of the 2D display, it has been criticized that it was designed in such a way that one must divide attention between street events and the display. Because the display is placed relatively far down in the field of vision, simultaneous perception seems impossible, and the placement was considered unpleasant in contrast to the 3D display.
Speed Control
Statements in this category are related to speed and its control (i.e., how well the speed could be detected and maintained). The keywords were “speed,” “optimal,” “ideal,” “hold,” and “recognize.” A total of 36 (35%) statements could be counted in the category of speed control, of which 22 (61.1%) are negative and 14 (38.8%) are positive. In addition, in this category, more statements are related to the 3D representation (22; 61.1%), of which 17 (77.3%) are negative and 5 (22.7%) are positive. The 2D statements are in the ratio of 9 (64.3%) negative to 5 (35.7%) positive. Both ratios can be interpreted more clearly due to their clear distribution. In the case of 3D, at higher speeds, the problem arises that the assessment of the current speed and its regulation is less precise. The bar becomes longer due to the depth of the room, which reduces the view of the ideal area. As a result, the “ideal speed is harder to hold” and “one can recognize speeds worse.” In 2D, all negative statements indicated that it was difficult “to keep the bar exactly in the middle.” However, the advantages lie in its simplicity and intuitive handling. It becomes clear that the 2D interface is preferred because of its simplicity, but the appearance of 3D is still mentioned positively.
Nonspecific Statements
Statements that did not refer to a specific aspect named above were assigned to this category. Primarily, general statements were included. The keywords were “better,” “prefer,” and “choose.”
This category is used to classify all statements that do not address the specific aspects (speed, display, and presentation, or shortcomings of these). Because there are only a few statements, all are presented. The statement about 3D was that “3D is better,” which was evaluated as positive. With 2D, no essential aspects were emphasized, but it was judged to be more pleasant overall. In addition, 2D was “to be preferred,” “the one to choose,” and “at least as efficient.” As a disadvantage, a participant mentioned that 2D “is useless and boring nowadays.”
Discussion
As described in the introduction, the objective of this paper was to investigate whether the ecological interface should be augmented, or whether a contact-analog visualization tradeoff exists due to human perception and cognition. The driving data, questionnaire data, and qualitative content analysis demonstrate that a tradeoff is indeed connected to the contact-analog visualization.
While 3D is considered more innovative and better integrated, the 3D version also better helps keeping the lane, as the paper’s introduction predicted an increase in performance when visualizing the boundaries of acceptable performance. Making the position of the car on the lane visually explicit enhances lane-keeping, which is based on the visually perceivable street boundaries. Therefore, if we desire existing boundaries (e.g., lane markings) to guide behavior, the behavioral indicator should be visualized in a contact-analog manner.
However, the augmentation also comes with some shortcomings, which can be explained by the dimensional representation. As the driving data reveal, the precision at which the driver follows the recommended speed is comparably low. It appears that the augmentation leads to a visual distortion that is large enough to cause this difference. Therefore, follow up studies should investigate which specific facet of depth perception, named in the paper’s introduction, is accountable for this effect. On the other hand, the augmentation and lane-keeping relation might be so salient that it draws attention to it.
Moreover, while 3D appears more satisfying, the questionnaire data and qualitative content analysis indicate that 3D is rated as more disturbing, demanding, and distracting. This is an important finding for further research. The paper at hand focuses on analyzing the contact-analog visualization tradeoff under predictable driving conditions. However, the qualitative results indicate the possibility of cognitive capture which might be a problem when something unexpected happens. Therefore, situations with unexpected events like animals crossing the street should be generated, to test whether objects in the periphery are more likely to be overseen when using the contact-analog visualization.
Finally, analyzing the cognitive demand caused by both versions of the interface, we found that no significant difference exists. The ecological interface approach works with both interfaces, as the possibility for direct perception-action mapping is given either way. Thus, we conclude that no extra value for speed control is provided in the dimensional representation, only that one can use the environment’s existing boundaries in this way. Based on this insight, one could advise on practical applications as follows:
Two dimensional displays should be used for use cases where a more precise control of behavior is needed and whenever the boundaries of acceptable performance are artificial and thus have no reference to the environment. Examples could be:
Speed → A fixed 2D representation should be superior. As this paper shows, the 2D representation enables a more precise control of behavior.
Cruise control → A fixed 2D representation should be superior. The artificial boundary of a set speed has no reference to the environment.
In contrast, 3D displays should be used for use cases where boundaries of acceptable performance have references to the environment. Examples could be:
Lane stability → A contact-analog 3D representation should be superior. As this paper shows, the 3D representation enables a more precise lane keeping behavior.
Following distance → A contact-analog 3D representation should be superior. The preceding car constitutes a boundary which can be used for direct perception-action coupling.
Adaptive Cruise Control → A contact-analog 3D representation should be superior. Preceding cars constitute boundaries which should be used for direct perception-action coupling.
In sum, we argue that, if we want artificial boundaries to guide behavior, the behavioral indicator should be visualized in a 2D manner, which is less prone to optical difficulties and thus enables a more precise control. However, if boundaries naturally exist in the environment, the contact-analog visualization might be superior because these boundaries then directly can be used to guide behavior. Moreover, the augmented visualization might be beneficial to adapt to unforeseen situations that are not anticipated by the interface but constitute a boundary of acceptable performance. Therefore, further research should examine such use cases in more detail.
Key Points
Ecological interfaces might make use of contact-analog visualizations.
Drivers prefer a 2D visualization because they perceive it as less disturbing.
Two dimensional visualizations in the HUD enable a more precise control behavior.
Contact-analog visualizations might make use of existing boundaries in the environment.
Footnotes
Appendix A: Instructions 3D
You are driving on a country road. Please follow the speed limits:
Your current speed is represented by a blue bar in front of your vehicle. This bar adapts to your current speed. This means that the blue bar becomes longer when you drive faster. A white bracket on the road indicates the target speed at any time. If the blue bar does not reach the white bracket, you should drive faster to reach the ideal speed. If, on the other hand, the blue bar exceeds the white bracket, you should drive slower to reach the ideal speed again. Ideally, the blue bar should end exactly in the middle of the white bracket.
Appendix B: Instructions 2D
You are driving on a country road. Please follow the speed limits:
Your current speed is represented by a blue bar in front of your vehicle. This bar adapts to your current speed. This means that the blue bar becomes longer when you drive faster. A white bracket indicates the target speed at any time. If the blue bar does not reach the white bracket, you should drive faster to reach the ideal speed. If, on the other hand, the blue bar exceeds the white bracket, you should drive slower to reach the ideal speed again. Ideally, the blue bar should end exactly in the middle of the white bracket.
Acknowledgments
Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - 227198829/GRK1931.
Author Biographies
Frederik Schewe is a PhD student in the Department of Engineering and Traffic Psychology at the Technische Universität Braunschweig. He received his master’s degree in psychology from the Technische Universität Braunschweig, Germany, in 2017.
Mark Vollrath is head of the Department of Engineering and Traffic Psychology at the Technische Universität Braunschweig. He habilitated in psychology at the University of Würzburg, Germany, in 2001.