1 Introduction

Augmented reality (AR) has long been a promising technology for tourism applications. While early systems were mostly different kinds of AR tourist guides with points of interest [1], the latest advancements in AR technology have made the development of more advanced AR applications feasible [2]. One interesting direction is related to using large 3D models such as models of entire buildings and letting the user walk freely inside the model and examine it using AR. Many important tourist attractions have already been modeled in 3D, and a model is thus readily available. This kind of feature is already available, for example, in certain real estate AR applications (e.g. [3]), and interiors of ancient buildings can be examined in some archaeological tours with AR glasses (e.g. [4]). However, this approach is still less often studied in the field of tourism. Naturally, virtual reality technologies could also be used for visualizing interiors, however, it may be difficult to provide virtual reality equipment for tourists, for example, when the attraction is located outdoors.

The current paper describes our first experiment with an AR application, which allows its users to walk freely inside a 3D model of a tourist attraction and to view the model by using tablet-based AR. A user evaluation was arranged with nine participants. The aim of the evaluation was to understand, whether the current approach of using AR technology in tourism can result in positive user experiences. In addition, the goal was to study, whether the users can successfully form a spatial situation model based on viewing the model. Finally, we were also interested in whether the users experience spatial presence while viewing the 3D model with tablet-based AR.

2 AR Implementation of Mannerheim’s Saloon Car

Mannerheim’s Saloon Car is a historical train carriage that was used by field marshal and president of Finland Carl Gustaf Mannerheim during World War II, when it housed important historical events. The saloon car is nowadays a tourist attraction located next to the railway station in Mikkeli, Finland. Because of the vulnerability of the attraction, the interior of the carriage can be visited only once a year on a festival day. A detailed 3D model of the saloon car was created based on lessons learned from previous work [5, 6] regarding the salience of visual cues in 3D models. The textures were closely modeled based on photos, and the model included furniture as well as several detailed objects including newspapers, maps, decorations, and Mannerheim’s personal items.

The augmented reality application for the saloon car was created using the Tarina augmented reality platform [7] developed by CTRL Reality. Using the platform, the 3D model of the saloon car was anchored to a location right next to the real saloon car (Fig. 1). Interaction was quite simple: the user saw the real size 3D model by looking through the tablet into the direction of the location, where the model was anchored. The model remained stationary, and the user could walk freely inside the 3D model, also through walls and furniture. The user could scan the 3D model by moving the tablet in his or her hands, or by moving or rotating her/himself, while looking through the tablet.

Fig. 1.
figure 1

AR implementation of Mannerheim’s Saloon Car displayed next to the real saloon car (left). Screenshot from actual user evaluation (right).

3 Methods

3.1 Participants

Nine Finnish participants (six males and three females; mean age 34.3 years, range 22–61 years) participated in the user evaluation. They rated their previous experience of using information technology as relatively high (mean 3.8 on a 1–5 scale), but their previous experience of AR applications as low (mean 2.0 on a 1–5 scale). They also rated both their interest in AR applications (mean 4.0 on a 1–5 scale) and their interest in historical tourist attractions (mean 4.2 on a 1–5 scale) as high. The participants received a small merchandise as a reward for their participation.

3.2 Procedure, Materials, and Equipment

The participants were recruited near the real saloon car at the Mikkeli railway station with the goal of carrying out the evaluation with participants who represent the target group of the study (tourists, who are interested in the saloon car). Indeed, a clear majority of the participants showed spontaneous interest towards the saloon car before they were recruited. The rest were asked whether they were interested. In all the evaluations, three researchers participated in the implementation of the study: one was recruiting and instructing participants, one taking care of the AR application and technology, and one taking care of the post-test questionnaire. All the evaluations were carried out on a single sunny, hot, and slightly windy summer day. Most of the time, the railway station area was nearly empty, but some of the evaluations were witnessed by other persons.

The participants’ task was to examine the 3D model of the saloon car using the tablet-based AR system. The participants were told that this can be achieved by looking through and moving the tablet and walking inside the model. The participants started using the system, and the test ended when the participant had walked through the whole carriage at least once and examined its objects using the AR system.

In the post-test questionnaire, a short version of Hassenzahl’s AttrakDiff2 user experience questionnaire exactly as used in [8] was used for evaluating user experience using eight scales. The four-item scale from the MEC-SPQ spatial presence questionnaire [9] was used to study the formation of a spatial situation model and the eight-scale Spatial Presence Experience Scale SPES [10] was used for studying spatial presence. The scales were translated literally from English to Finnish so that the AR saloon car was defined as the object of evaluation.

A Samsung S7 tablet with 6 Gb of RAM memory and an 11″ display with a resolution of 1600 × 2560 pixels was used by the participants to view Mannerheim’s saloon car in AR. The system was able to run the AR application and display the 3D model smoothly. The participants used another tablet to respond to the post-test questionnaire.

3.3 Data Analysis

Due to the distribution-free nature of the data, Friedman’s rank tests were used to compare the ratings of the different user experience and spatial experience dimensions for significant differences, and Wilcoxon’s matched pairs signed ranks tests were used in pairwise comparisons. Internal reliabilities were studied using Cronbach’s Alphas.

4 Results

Mean ratings and standard errors of the means for the user experience dimensions and the spatial experience dimensions are presented in Table 1 below. Larger scores indicate better user experience or spatial experience.

Table 1. Mean ratings and standard errors of the means for the different user experience and spatial experience dimensions

The statistical analysis showed that there were no statistically significant differences between the different user experience dimensions. In contrast, there were significant differences between the three ratings related to spatial experience χF2 (3) = 6.4, p < .05. Spatial situation model received higher ratings than both spatial presence: self-location Z = 2.1, p < .05 and spatial presence: possible actions Z = 2.2, p < .05. Internal reliabilities were on an acceptable level for all the components: spatial situation model α = .77; spatial presence: self-location α = .95; spatial presence: possible actions α = .90.

5 Discussion

The results of this first study indicated positive user experience for the tablet-based AR implementation of Mannerheim’s Saloon Car on all main dimensions of the AttrakDiff2 scale: pragmatic quality, hedonic quality (identification and stimulation), and attractiveness. In addition, the results indicated that the participants could form a precise mental spatial situation model of the saloon car (the floor plan and the sizes of the different rooms etc.). They also experienced relatively high spatial presence, even though they just used a tablet to view the 3D model instead of immersive technologies. It should also be noted that the tablet-based system was used on a sunny day, and the participants had low previous experience of AR systems, but the results were still good.

Some initial challenges related to the usage of the system were identified based on on-site observations, screen recordings of the tests, and the participants’ post-test comments. A few users became briefly disoriented during the use of the system, which was mostly related to walking through inside or outside walls of the carriage instead of using doors and corridors. Two users started walking backwards, which also caused some disorientation. One participant also accidentally blocked the camera of the tablet briefly with his finger. However, the most important activities such as walking along the main corridor of the carriage and looking into the different rooms of the carriage could be performed well. Suggestions for improvement included more objects and interactivity with objects. Interactivity is an important challenge for the future, as current AR platforms do not do an especially good job in providing support for interactivity within 3D models in a tablet-based AR application.

Overall, the results suggest that the current approach, in which the users can walk inside the 3D model of a tourist attraction using a tablet-based AR application, can evoke a good user experience and is a noteworthy option to consider, when tourists cannot be let to enter the attraction itself. The AR approach could be especially useful in situations, in which virtual reality experiences cannot be effectively provided for the users, such as in the context of different unstaffed attractions located outdoors. Clearly, however, the findings from this experiment need to be confirmed by further studies.