Keywords

1 Introduction

The rapid advance of Information and Communication Technologies (ICT) in the last decade has allowed their growing integration in all areas of society. Education has not escaped this, especially in the field of architecture and architectural representation, where the use of new technologies such as Augmented Reality (AR) or Virtual (VR) has led to a true revolution in the classroom [1]. Among its benefits is its contribution to improving motivation, satisfaction level and academic performance [2, 3]. However, the investigation of its impact on teaching is still incipient and the results are scarce, especially in the field of spatial perception, where immersive VR, turned into a powerful tool for realistic simulation of an architectural space, could be an effective tool in the training and development of this cognitive ability. Thus, the appearance of Immersive Virtual Environments (IVEs) provides new opportunities to create realistic experiences in spaces that do not exist, turned into digital twins that, as a test bed, must allow to simulate the physical conditions of a real architectural environment.

The proposed methodological proposal is based on the hypothesis that a correct perception of the environment should allow accurate estimates of: (a) the measurements (height, width, length) of an architectural space; (b) the configuration or relative position of a space or an object with respect to another; (c) attributes such as the finish and quality of each of the perceived spaces. It is assumed that these estimates must allow us to move and perform tasks efficiently and effectively in a virtual environment, helping decision-making, while the user will be able to subjectively perceive the quality of the virtual space in a similar way to what he would do in a real environment (See Fig. 1).

Fig. 1.
figure 1

Source: prepared by the author

Comparison of the real space (left) with the recreated virtual environment (right).

Full size image

2 Background

The way in which people perceive a real space as opposed to their perception in a virtual environment has aroused growing interest, parallel to the development of technology [4]. In Henry and Furness [5], a similar experiment similar to the one proposed in this writing can be found, where the way in which people perceive real and virtual spaces is compared. In it, increasingly immersive virtual representation techniques are used (although currently obsolete), and the participants carry out tasks of dimensioning, orientation, and qualitative evaluation of the space. The results then suggested that the subjects underestimated the dimensions of the space represented virtually in comparison with those realized in the real space.

Numerous studies can be found in the same line of underestimation of the perceived distances in IVEs [6, 7]. In general, they focus on spatial perception understood as the ability to measure distances between the subject and an object by comparing the two spaces. In most cases, they report significant errors in the perception of distance, mostly due to factors such as the focal length of the device and the quality or resolution of the image [6, 8, 9]. Other studies aim to evaluate how spatial perception helps to prioritize and process the information received from a given environment [10] or explore the relationship between the perceived size within a space based on its shape [11,12,13]. In the same way, certain authors evaluate the user’s performance by comparison with different environments [14, 15], even using study methods similar to the one proposed in this writing [7, 16]. All of them have contributed to a better understanding of human perception in EVIs, although the disparity of reported results does not help to trust the use of VR as a research tool for cognitive experiments. Furthermore, focused on the precision of the evaluation of the perceived distances, the user subjective perception of the space is not addressed.

In this context, it therefore seems appropriate to propose a more extensive evaluation of spatial perception in virtual environments, not only in relation to its physical characteristics, its measurements, its shapes, our position, and movement [17], but also in relation to the subjective perception of the visualized space.

Fig. 2.
figure 2

Source: prepared by the author.

Example of different representations of a virtual environment created as a digital twin from a real existing space. In this case, for materials and lighting analysis.

Full size image

3 Methodological Proposal

The experience will be carried out in three of the adjoining classrooms of the EPSEB-UPC, whose configuration has been simulated as a digital twin that will serve as a test base for this and other projects (see Fig. 2).

First, two random groups will be selected, formed by volunteers from the degree of architecture and technical architecture, who will carry out a PRE-TEST of spatial skills, where equivalence between the groups will be ensured, before carrying out the training, through a variance analysis. The confidence interval will be set to 0.05. Due to the large number of existing tests [18], this preliminary test to which the two groups will be subjected is inspired by classic tests such as the Mental Rotation Test (MRT) [19] and the Differential Aptitude Test - Spatial Relations Subset (DAT–SR) [20].

Once the two equivalent groups have been formed, a series of activities will be carried out in a real environment (control group) and the same activities in a virtual environment (experimental group) in which three components will be evaluated:

  1. 1)

    SIZE and SHAPE: The most basic attribute of space is its shape and size. The simplest way to measure people's perception of room sizes is to ask them directly but very few people feel they can be accurate in expressing distances using a metric system, such as feet or meters, because metric distances are not immediately intuitive. By contrast, most people are quite comfortable expressing distances as a function of other distances. For this reason, most psychophysical studies use a “ratio estimate technique”, such as Sadalla’s room size experiment [11]. Although this technique could be helpful, it leaves room for quite a bit of imprecision, and certainly more than can be tolerated for this kind of study. The option of asking directly about the perceived distances of the participants has been finally chosen, since the study will be carried out by students in the field of architecture, accustomed to estimating dimensions in real spaces.

  2. 2)

    SPATIAL ORIENTATION: The way one feels in a particular space is partly affected by the perception one has of that space’s location within the overall project. A good way to measure the accuracy of people’s perception of their place in the overall layout of a virtual environment could be to study an sketch of the cognitive map they form of these spaces [21]. But the transcription from a 3d Cognitive map to a 2D plan view requires a particular kind of skill without which the interpretation of the sketches become meaningless. Furthermore, the sketch technique is difficult to analyze. In this approach, we’ll use the “point to direction” technique [5]. Participants move about a space and at specific moments, they’ll be asked to stop and determine in what position he is, and the relative position of characteristic objects located in the scene that they have seen previously.

  3. 3)

    INDIVIDUAL SENSATION of PERCEIVED SPACE: While the importance of how a place feels is the overall goal of a successful simulation, it’s perhaps the most difficult factor to measure since there are no objective variables to evaluate the quality of a space. In our case, a checklist of adjectives will be used, which is the most common measure for tasks that require quantitative estimates of people's spatial descriptions [22, 23]. This technique consists of administering a list of bi-polar adjectives. The participant selects all the adjectives which apply to the tested space. The results of the virtual environments can be compared to those for the real place. Their level of correlation becomes the measure of “goodness” of representativeness of the simulation condition.

Participants from control group will be requested to make a tour through rooms 1,2,3. During this tour they’ll be asked about component 1. (size and shape). Once they’ve finished, they’ll be asked about component 2 (Spatial orientation). And finally they’ll fill the questionnaire C (Individual sensation of perceived space) (see Fig. 3).

The tests will minimize the participation of other cognitive functions as much as possible. Parameters 1 and 2 will be evaluated by “in situ” test. For the evaluation of component 3, a specific questionnaire will be carried out at the end of the activity, which will also include the evaluation of the usability of the system used. Evaluating then its efficiency, effectiveness, and degree of user satisfaction.

Once results will be obtained, independently, their interpretation is complex. To provide a clearer interpretation and allow the information to be presented in a brief and concise manner, there is a need to group these responses. Because of, it is necessary to construct a composite indicator of spatial perception. This “quality” indicator cannot be measured in units but will allow comparisons between participants and groups, and correlate it with other indicators, such as age or gender for a better understanding of factors affecting spatial perception in virtual environments.

To achieve this interpretation, we’ll use Principal Component Analysis (PCA). Once major components (Size and shape, orientation, and individual sensation) and contribution rates will be estimated, each participant we’ll be rated according to an index derived from a general expression that weighs the scores for each principal component to the square root of the variance as follows.

Fig. 3.
figure 3

Source: prepared by the author

Up: proposed route to be carried out by the control and experimental groups.

Full size image
$$ I_{mj} = \frac{{\mathop \sum \nolimits_{i = 1}^{r} Z_{rj} .\sqrt {\lambda_{r} } }}{{\mathop \sum \nolimits_{i = 1}^{r} \sqrt {\lambda_{r} } }} $$
(1)

In the Eq. (1),

Imj represents the composite indicator to be achieved (efficiency, satisfaction, effectiveness, etc.) for each j-th participant.

Zrj score is the r-th component (factor) for the j-th participant.

λ is the square root of the “eigenvalue” r for that component, ensuring that the components with higher explained variance have a greater weight in the index construction.

The proposed methodological scheme is fully shown below (see Fig. 4).

Fig. 4.
figure 4

Source: prepared by the author

Proposed methodological scheme.

Full size image

4 Conclusions

The results of the three tests described should serve as the basis for comparing an existing space with a realistic digitally simulated immersive environment. A high correlation between the test results should indicate that these virtual interfaces are accurate tools for representing real environments, concluding that they could be used as architectural simulation tools. If not, the results could help to detect certain gaps and areas for improvement in VR technology.

The construction of an indicator of spatial perception could represent a useful approach for the study of this cognitive ability and could help to draw conclusions in a more objective way about the factors that affect this ability. At the same time, it will allow comparison with other factors such as age and sex, and even with other similar experiences, which could make progress in this field of research possible.