fun.tast.tisch. is an interactive tabletop system used in neurorehabilitation. It consists of several exercises to be interacted with by therapists and patients collaboratively. The user interface combines di erent interaction paradigms and tangible objects to facilitate therapeutic concepts. The acceptance of the system and its user interface relies on therapeutic aspects but also on straightforward interaction, usability and a positive user experience. A study with nontarget-group users was conducted, rst, because earlier studies revealed that the target groups are not fully impartial regarding perceived usability/user experience. Second, interaction combining a touch-based interface with tangible elements and system feedback on di erent sensory channels involves usability/UX challenges that should be investigated in a non-critical situation. This paper describes selected fun.tast.tisch. modules, especially those involving tangible objects and summarizes the results of a related study.
H.5.2 [Input devices and strategies]: User Interfaces; H.5.2 [Interaction styles]: User Interfaces; H.5.2 [Usercentered design]: User Interfaces
The tabletop system fun.tast.tisch. was developed as a
supplement to conventional therapy in neuro-rehabilitation
[
Most patients working with fun.tast.tisch. su ered acquired
brain injury (often caused by stroke or accident) and
incurred cognitive and/or motor impairment. Conventional
therapy aims at training, e.g., attention, memory,
visuospatial abilities or motor skills; thus related therapeutic
concepts are the basis for the di erent fun.tast.tisch. modules
as well. The development process is accompanied by
evaluation along di erent dimensions. First, there have been two
longer-term studies in a rehabilitation hospital (see [
The study summarized in this paper had di erent objectives. First, we wanted to test new tangible elements in a non-clinical context. Second, we felt during earlier studies that therapists and patients naturally focus on therapeutic aspects (we observed that their perception of usability and UX tends to be biased by issues related to the underlying therapeutic concepts). Thus, we used a setting with a comparatively high number of non-target-group participants and tested 7 modules (3 had not been tested in a larger-scale study before, 4 involved di erent tangible objects).
This section provides an overview on related approaches in the area of virtual rehabilitation, focusing on the integration of tangible elements.
De la Gu a et al. [
Sharlin et al. [
Rand et al. [
Wilson et al. [
Alamri et al. [
THE FUN.TAST.TISCH. UI fun.tast.tisch. is an interactive tabletop system, thus a major way to interact with the UI is via touch. Our tabletop technology uses an optical display, thus it is also possible to recognize physical objects placed on the screen. fun.tast.tisch. involves several tangible elements as interaction with physical objects is well suitable for e.g., training of motor or spatial skills. fun.tast.tisch. does not involve further interaction with conventional means (such as keyboard or mouse). This section describes brie y the di erent interaction modalities. 3.1
The navigation through the main portal, including users' login and registration, and the selection and initial con guration of the modules, is done via touch. As entering textual information is a bit inconvenient via touch, we aimed at reducing the amount of necessary textual input to the minimum possible degree (the registration process is the only part of the system that requires it). The system allows for \sessions" that therapists and patients are assigned to. The session information is needed, as the system also provides a statistics feature. Therapists' and patients' registration can be done using a virtual keyboard, login is however intended to be implemented using tagged cards to be placed on the display. It is also possible to work in an anonymous mode that does not require registration (this however means that sessions later cannot be associated with the users). For the therapist's tasks, the main part of the interaction is via simple touch-based selection. For patients (who usually don't navigate through the settings dialogues but solve exercises), there are two more ways of interacting via touch, a twohanded touch, where two di erent target spots have to be touched at the same time (this is actually relevant for therapists also), and a delayed touch that has to be held for a longer time to select an object on the screen. The latter was introduced for two main reasons: rst, the exercises using it require the user to purposefully select objects (e.g., if multiple objects are available and the user has to choose the correct one) and we aimed at preventing a trial and error approach where users just wipe over the display or touch randomly. Second, we wanted to avoid unintended selections with arms or elbows. A delayed touch starts an animation that is nished when the required duration is reached (see Figure 1 { the green circle around the touch position indicates that the delayed touch has been started).
Several fun.tast.tisch. exercises, especially those training
motor and spatial skills, bene t from the integration of
tangible objects (see Section 4 for details). Generally, our UI
involves two di erent kinds of tangible elements: rst,
objects that are recognized via tags, and second, non-tagged
objects that are either recognized by the system based on
their shape or do not require to be recognized. Tagging
objects is the simplest way to have them recognized, which is
why we chose this option in all cases where it was possible.
However, some aspects of our interaction design did not
allow for the objects to be tagged, e.g., because some objects
need to be ipped over by the patients and be used from
both sides. We decided not to tag both sides of ( at) objects
for aesthetic reasons, and because the semi-transparency of
these objects should be exploited for a certain kind of visual
feedback (see Figure 2 [
The material of all objects was chosen carefully based on di erent considerations. Some physical interaction elements needed to be semi-transparent, others needed to be fully transparent (e.g., in case the system should not recognize them). The modules that allow for a multi-patient setting needed objects of di erent patients to be distinguishable, which was solved by di erent colors. Further, all objects needed to be easily clean- and disinfectable, well graspable (these are domain-speci c requirements) and needed to be solid enough not to break when dropped. After tests with di erent kinds of materials, we arrived at i) objects made of acrylic glass, individually designed for fun.tast.tisch., and ii) 3D printed ones. Figure 3 shows the di erent kinds of physical objects currently used within fun.tast.tisch.
This section presents the 7 modules that have been tested
regarding usability/UX in the study with non-target-group
users. One module (Tangram) has been exhaustively
evaluated earlier [
This module (see [
Where's the Food? (see a picture in [
Window Washing (see Figure 5) trains motor skills. The table shows a picture covered with digital dirt that has to be \wiped free" by the user. The therapist can choose between three picture sizes (depending on the user's reachable screen areas). The patient uses a physical object (or alternatively a common towel) that can be xed with his/her hand to wipe away the dirt (as shown in Figure 5). The object is transparent because it should not be recognized by the system (the dirt is removed based on the shape of the user's hand). 4.4
This module (see Figure 1) displays to the user a number of lled colored circles, two of each color. This module uses two-handed and delayed touch (see Section 3.1). The user has to nd the two circles belonging together and touch them (one with each hand) at the same time, which will make them disappear. The process has to be repeated until no circles remain. The therapist can set the number of pairs that should be displayed and choose whether to use the colored circles or pictograms. If the user cannot use both hands, the module can be con gured to be operated with one hand only (via delayed touch, the two items have to be touched one after the other within a certain time span). 4.5
Spot the Di erence [
This module (see Figure 6) trains spatial cognition skills. The therapist places a set of physical objects on the display, the table then generates di erent up to 4 three-dimensional models that show the objects from di erent perspectives (the patient has to choose the correct one). The therapist
This module (see Figure 6) trains spatial construction skills. The therapist places a set of physical objects on a certain area of the display, the patient then has to rebuild the setting with a second set of identical objects. The therapist receives an overlay picture generated by the system showing the footprints of both sets of objects and can decide how well the patient has solved the task. The module uses the same objects as the Spatial Cognition module, thus the therapist can choose between simple and complex ones. 5.
This section describes the general aims and setup of the study, the applied method and participants. 5.1
It was our aim to evaluate selected modules along general usability and UX related dimensions, intentedly not speci c to the domain of neuro-rehabilitation. 85 probands participated (39 male, 44 female, 2 unknown). They were between 10 and 54 years old (x = 21:87). Most of them showed high a nity towards technology, 89% stated to own a smart phone and 88% to be experienced with touch devices and use them on a daily basis (11 had little or no experience). They participated voluntarily during a university open house. 5.2
We followed a task-based user test approach. The tasks
to be done by the participants aimed at covering all
possible user interactions, including use of tangible objects.
Every participant did one or more tasks and answered a
web-based questionnaire afterwards. The questionnaire
contained about 230 questions in total, however, the
participants received a personalized set of questions (in most cases
between 30 and 50) based on the modules they tested. As
fun.tast.tisch. is designed for therapy settings, we split up
the tasks into subtasks for therapists and for patients. Each
task required a setting with one participant to take over
the therapist's role and at least one \patient". We aimed
at getting an overview on the perceived usability and UX
of the whole UI, but also speci c to each of the evaluated
modules. In total, we received results for 166 role/module
combinations: 16 \therapists" and 25 "patients" for the
Shopping module, 7 \therapists" and 7 "patients" for Where's the
Food?, 9 \therapists" and 13 "patients" for Window Washing,
9 \therapists" and 11 "patients" for Match the Pairs, 6
\therapists" and 9 "patients" for Spot the Di erence, 14
\therapists" and 18 "patients" for Spatial Cognition, and 9
\therapists" and 12 "patients" for Spatial Construction. Based on
the classi cation of usability-related criteria by [
This section provides a concise overview on the study's results, mainly related to interaction.
The e ectiveness of the system was partly investigated by asking whether the cooperation between ( ctitious) therapist and patient(s) worked well, which most participants unanimously a rmed. For therapists, other questions dealt with control of the work ow, e.g., related to starting/pausing modules or changing settings (e.g., the level of di culty). Most cases where problems were reported were related to insu cient task descriptions (not relevant for application in real settings) and minor UI issues (like e.g., a misleading button label). Patients' questions were related to the exercises themsevles, e.g., whether selecting or moving elements worked well. The delayed touch interaction that turned out to be critical in earlier tests worked and was understood well. Nevertheless, this kind of interaction must be explained beforehand as it is still not fully intuitive, although we tried to design the animation to indicate that the touch needs to be held for a longer time.We asked patients if the di erent perspectives in Spatial Cognition were clearly recognizable, i.e., di er su ciently (17 of 25 found this to be \very good" or \good'). Regarding the handling of the physical objects we found out that participants liked them and understood how they should be used well. At the Shopping module, all but one of 25 patients stated to have known how to use the coins. 23 of 25 found the shape of the coins and properties of the material \very good" or \good". The coins were e.g., described as follows: \perfect size", \easy to handle", \easy to grasp", \funny", \ ts well in the hand", \compact and handy". The only negative description was \a bit too small". 8 of 14 found the shape and properties of the objects in Spatial Cognition \very good", 4 \good". Regarding Spatial Construction, all patients and all but one therapists declared that they knew how they had to place objects on the table and all but one liked the shape and material.
In order to identify potential safety issues, participants e.g., were asked whether anything unexpected occurred during the training. The results show a higher number of unexpected events mainly in the Shopping module. However, the main situation participants described as \unexpected" was the \distraction" game that started between memorizing goods and picking them from the shop in order to increase the memory span. Although this behavior was of course unexpected for rst time users, it is desired. In summary, we identi ed minor safety issues most of which originate in technical issues that partly have been solved meanwhile.
The utility criterion is concerned with the individual user (and his/her ability to use a system in the ways he/she wants to). Accordingly, we asked utility questions in cases where a task could be performed in more than one way. E.g., at the Shopping module, users might have forgotten goods they should have memorized. From a utility perspective, it is of interest if the system provides su cient means to help the user achieve the task. 5 of 25 patients made use of the physical shopping list and all of them stated that they understood how it could be used. However, 2 participants initially placed the physical list on the table upside-down. Another utility example is the therapist's menu: all exercises can be ended by using this menu that is not intended to be operated by patients. We also had to make sure this menu is not activated unintendedly (as it is located near the therapist's sitting position). Thus, users have to perform a two-handed touch (on two corners of the table) to activate it. 7 of 9 therapists knew how to use the menu at the Window Washing module, 8 of 9 at Match the Pairs, 11 of 14 at Spatial Cognition, and 7 of 9 at Spatial Construction.
Regarding learnability, we aimed at nding out how fast the participants understood how the exercises worked. With only few exceptions, all therapists and patients stated to have understood how the exercise worked \very fast (was self-explanatory)" or \fast" for all modules. No participant selected \very slow (was confusing)" or \slow". Further, we asked how well participants got along with the overall system (navigation, etc.) to nd out how fast they could work with it after a relatively short period of time. On a scale from \very well" (1) to \very badly" (5) , 35 patients rated 1, 20 chose 2, 4 chose 3, 1 chose 4 and 0 chose 5. 26 therapists rated 1, 17 chose 2, 3 chose 3, 0 chose 4 or 5.
Summarizing usability-related ndings, no critical usability threats but a number of smaller issues were identi ed. The interaction with tangible objects was perceived exceptionally positive although most required a short introduction on how they should be used to interact with the system. After the introduction, there were no complaints about handling the objects or understanding how they should be used. 6.2
Regarding UX, participants had to describe how well four of the aforementioned UX adjectives described the interaction with the system at the module level. They were asked whether they found the module to be \fun", \entertaining", \motivating" and/or \aesthetically pleasing" (each on a 5item Likert-scale). We used this reduced set of adjectives because not all of the other adjectives would have matched all kinds of exercises. Another aspect evaluated at the module level was the perception of feedback. Patients were asked whether they noticed feedback and if so, if they understood it and to what extent they perceived it as \motivating". The results revealed that in all modules the system's feedback was perceived as (very) motivating by the majority of the participants (16 of 19, 6 of 7, 9 of 9, 5 of 6, 6 of 7, 12 of 14 and 10 of 10). Additionally, both groups were asked how they perceived the overall interaction with fun.tast.tisch., using a larger set of adjectives. As shown in Figure 7, the attribute \fun" was most often (by 63) fully agreed with (additionally, no participant selected \disagree" or \fully disagree"). Given the study was conducted with non-target-group users, this is a highly desirable outcome. Other studies have already shown that for the actual target group, mainly patients, the system is perceived as motivating and helpful. Non-targetgroup users are naturally less in need of motivation for therapeutic exercises and might thus have rated attributes like \satisfying", \motivating" or \helpful" less important.
We have described the di erent interaction paradigms of
fun.tast.tisch. and selected modules and presented the
results of a study conducted to evaluate the usability and
UX with non-target-group users. Even though we
intentionally excluded the actual target groups from this study,
they play the most important role in the overall evaluation of
fun.tast.tisch. The system has been repeatedly tested with
real users, e.g., in longer-term evaluations at a rehabilitation
hospital [
fun.tast.tisch. acknowledges nancial support within the COIN (Cooperation and Innovation) program, managed by the Austrian Research Promotion Agency, funded by the Federal Ministry for Transport, Innovation and Technology and the Federal Ministry of Economy, Family and Youth. Project partners are LIFEtool, ARTGROUP, UAS for Health Professions Upper Austria, UAS Upper Austria, Irene SchacherlHofer, softaware gmbh and TRANSPARENT DESIGN.