=Paper=
{{Paper
|id=Vol-2685/paper5
|storemode=property
|title=User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control for a Mobile Location-Based Augmented Reality Solution for Maintenance
|pdfUrl=https://ceur-ws.org/Vol-2685/paper5.pdf
|volume=Vol-2685
|authors=Moritz Quandt,David Hippert,Thies Beinke,Michael Freitag
|dblpUrl=https://dblp.org/rec/conf/ectel/QuandtHB020
}}
==User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control for a Mobile Location-Based Augmented Reality Solution for Maintenance==
https://ceur-ws.org/Vol-2685/paper5.pdf
H. Söbke, J. Baalsrud Hauge, M. Wolf, & F. Wehking (eds.):
Proceedings of DELbA 2020
Workshop on Designing and Facilitating Educational Location-based Applications
co-located with the
Fifteenth European Conference on Technology Enhanced Learning (EC-TEL 2020)
Heidelberg, Germany, Online, September 15, 2020
User-Centered Evaluation of the Learning Effects in the
Use of a 3D Gesture Control for a Mobile Location-Based
Augmented Reality Solution for Maintenance
Moritz Quandt1, David Hippert1, Thies Beinke1,
and Michael Freitag1,2
1 BIBA - Bremer Institut für Produktion und Logistik
at the University of Bremen, Bremen, Germany
2 Faculty of Production Engineering, University of Bremen, Bremen, Germany
qua@biba.uni-bremen.de
Abstract. Mobile Augmented Reality (AR) solutions are ascribed to a high po-
tential for location-based support in the work context. The technology enables
the insertion of virtual content directly into the working environment. The suc-
cessful introduction in practice of the developed solutions is highly dependent on
the acceptance of the end-users. Since there are no general design principles for
integrating novel forms of interaction and user interfaces into a three-dimensional
application environment, we apply user-centered evaluation methods. In this pa-
per, we investigate the learning effects of the users in handling a hand-based ges-
ture control using the example of an AR application to support the maintenance
processes of heating, air conditioning, and cooling systems. The users perform
five tasks in two successive test runs. Based on the processing times and the re-
quired interactions for each task, we can evaluate the applicability of the selected
interaction patterns for the respective task.The user study results show that users
learn to use hand-based gesture control in a short time. Especially when directly
manipulating virtual objects, the users quickly showed improvements regarding
processing time and number of interactions needed. In contrast, learning effects
in the use of the hand-gesture control do not become evident when performing
multi-step gestures without reference to the real environment. Since existing in-
teraction patterns do not necessarily achieve high user acceptance in this context,
user studies can provide valuable insights for the design of mobile location-based
AR solutions.
Keywords: Augmented reality, location-based information provision, 3D hand
gesture control.
Copyright © 2020 for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
Moritz Quandt, David Hippert, Thies Beinke, and Michael Freitag
1 Introduction
The reality-virtuality-continuum of Milgram and Kishino classifies Augmented Re-
ality (AR) as a technology that extends the real world with virtual content. While virtual
reality focuses on the complete immersion of the user in a virtual world, AR focuses on
the coexistence of real and virtual objects [1]. Azuma defines AR as the combination
of virtual reality and the real environment with partial superimposition, interaction in
real-time, and a three-dimensional (3D) relationship between virtual and real objects
[2]. For work process-integrated support and context-related information provision in
the work environment, AR technology is particularly suitable [3], [4]. In the field of
industrial applications, product design, manufacturing, assembly, maintenance, and
training are seen as the main application areas [5]. In the area of maintenance, several
studies have developed promising approaches to improve employees' performance in
the execution of technical maintenance tasks, for the training of employees to perform
maintenance tasks or to support the documentation of maintenance activities [6]. Espe-
cially in maintenance, the documentation and transfer of knowledge of experienced
service technicians play an important role. This know-how for the maintenance of ma-
chines and components is essential for the efficient processing of maintenance orders.
However, only a limited number of AR solutions to support service technicians with
location-based information in the work environment have already been used in practice
[6].
In order to increase the acceptance of AR-based solutions by users in practice, the
optimal interaction between humans and technology is the decisive criterion for the
development of AR-based assistance systems [7]. The high number of possible forms
of interaction, hardware configurations, and the possibility of addressing different
senses (visual, auditory, tactile) obstructs the development of generally applicable,
comprehensive design guidelines for AR applications [8]. As a basis for the develop-
ment of interactive AR systems, general requirements for the design of industrial AR
solutions [9] or dialogue principles, e.g. according to DIN 9241-110, are available. The
decision on the specific design of user interfaces and selecting suitable interaction pat-
terns depends on the individual application. Therefore, the involvement of the later end-
users is of great importance for the development of usable mobile AR solutions.
In [10], Quandt et al. presented a user study to evaluate the subjectively perceived
usability and the workload associated with the use of a location-based AR application
to support service technicians to conduct maintenance measures on complex heating,
air conditioning, and cooling systems on their work location. By taking user feedback
into account in the development of the presented AR solution, we improved usability.
In addition to the location-based support of service technicians in the work process,
which we tested with the system users based on usability and workload evaluation, this
article focuses on the learning effects of the users in handling the AR application. Es-
pecially concerning the used hand gesture-based user control, further research needs
have emerged. In the course of conducting the user study, we observed that users usu-
ally learn to use the 3D hand gesture control quickly. After this learning phase, users
become more confident in using hand gesture control. Consequently, an optimized de-
sign of interaction patterns to fulfill specific work tasks can be concluded. The learning
User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control
effects that occur when using hand gesture-based interaction will be examined in this
paper using the example of the AR application introduced.
Following the related work in chapter 2, we present the case study in chapter 3. In
chapter 4, we present and discuss the results of the user studies. The final chapter sum-
marizes the findings of the paper and provides an outlook on further research needs.
2 Related work
In their review, [11] examined the use of AR for industrial application scenarios. The
application areas of AR in maintenance deal with the training of employees to perform
maintenance tasks, process support for error prevention, maintenance of complex ma-
chines and compliance with safety guidelines, and the performance of maintenance ac-
tivities in hazardous environments. The systematic literature review by [6] provides the
state of the art in research on the use of augmented reality to support industrial mainte-
nance activities. The identified state of research includes AR-based assistance systems
in various application areas, such as aviation, plant maintenance, or mechanical mainte-
nance. [12] discuss remote support and work process support through virtual infor-
mation on maintenance objects as core applications for AR in maintenance. In various
studies considered for the development of solutions, the focus lies on the tracking pro-
cedures used, the mobile AR hardware used, or the interaction between humans and
technology. [13] identified great potential to use AR systems for service technicians'
training, with the possibility of AR to simulate real work situations. In this context, [14]
developed an AR-based learning platform that provides step-by-step instructions for
service technicians in the assembly and maintenance of industrial components and
plants. An instructor can use the live video image of the trainee to influence the task
execution. For location-based learning, [15] have developed an algorithm that identifies
real-life learning objects based on the learner's location and provides corresponding
learning content. Since many industrial applications require the users' indoor location,
the exact localization of the users is a central challenge in order to enable an accurate
superimposition of the virtual content. For this purpose, marker-based, SLAM (Simul-
taneous Localization And Mapping)-based and model-based tracking methods are used
in particular. In addition to achieving high accuracy, these methods need to be imple-
mented on mobile AR hardware [16].
Another central challenge in the development of mobile AR-based assistance sys-
tems is the interaction with virtual objects in the three-dimensional space. The use of
mobile AR hardware alters the requirements for the development of AR user interfaces
compared to classical WIMP (Windows, icons, menus, pointers) user interfaces of
desktop applications [8]. When using data glasses, hand gesture-based controls are in-
creasingly used. This type of human-technology interaction is particularly suitable for
AR applications for direct and, thus, intuitive interaction with virtual objects [17]. The
few formal evaluations of hand gesture-based interaction apply user-centered evalua-
tion methods, such as questionnaire-based evaluation of usability and acceptance [17],
or by recording and analyzing performance measures from user experiments [18].
Moritz Quandt, David Hippert, Thies Beinke, and Michael Freitag
To sum up, the challenges for introducing location-based mobile AR applications to
support work processes in maintenance are a context-based provision of information,
the reliable and accurate recognition of objects, and the use of appropriate kinds of
human-computer interaction. In this paper, we focus on the aspect of the experimental
testing of a hand gesture-based control system. With the results of the user study, we
plan to gain insights for the task-dependent selection of suitable interaction patterns.
3 Case Study
The maintenance of heating, air conditioning, and cooling systems in large infrastruc-
tures, such as department stores or airports, places high demands on the service techni-
cians' qualifications. The service technicians' work includes the orientation in the work-
ing environment, finding components, documenting measured values, detecting and re-
porting damages in the context of the maintenance measures carried out. Currently,
service technicians conduct maintenance activities with paper-based documents. The
service technicians usually carry a maintenance checklist for documentation purposes
and a revision plan that contains maintenance components listed in a floor plan of the
building. During the execution of maintenance tasks, the search for individual compo-
nents leads to a considerable loss of time. Due to the often missing documentation of
plan changes during component assembly, these search efforts occur. Therefore, up-
dated revision plans can contribute considerably to a more efficient work process. By
superimposing the virtual planning basis on the real objects, the trade-specific symbols
of the individual components can be displayed and manipulated directly in the field of
vision of the service technicians. This way, the service technicians both learn how to
work with digitized building data in the work process and contribute to an increased
efficiency in the maintenance process through the continuous actualization of the doc-
umentation. In this application case, the use of AR glasses offers the advantage of a
hands-free usage. Therefore, the technicians' ability carry out maintenance activities is
not restricted. Since the service technicians are working indoors, the orientation in the
work environment is based on room geometry, derived from the existing building plans.
For this purpose, an importing tool processes the existing revision plans for display
on the AR hardware. The importing tool transfers the plans to the mobile hardware
according to defined modeling conventions, which, for example, require the arrange-
ment of the room walls on one level of the plan. At the place of maintenance execution,
the AR system aligns the virtual revision plan with the real work environment. The user
supports the superimposition by setting a starting position that the system matches with
the respective revision plan. By moving objects installed at a different location than
specified, adding new objects, or deleting objects, the service technicians can directly
update the virtual revision plan. After completion of the maintenance task, an export
tool prepares the updated revision plans for a subsequent transfer to the order manage-
ment.
We conducted a user study to evaluate the subjectively perceived usability and the
workload associated with the use of the AR application. The results of this study were
User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control
presented in [10]. By taking user feedback into account in the development of the pre-
sented AR solution, we improved its usability. Especially concerning the used hand
gesture-based user control, further research needs have emerged. In the course of con-
ducting the user study, we observed that users usually learn to use the 3D hand gesture
control quickly. After this learning phase, users generally become more confident in
using hand gesture control, and conclusions can be drawn about the design of interac-
tion patterns to fulfill specific work tasks. The learning effects that occur when using
hand gesture-based interaction will be examined in this paper using the example of the
AR application presented.
4 User tests
At the time of our development of the presented AR application, Microsoft HoloLens™
best met the requirements of the application in the field of maintenance of heating, air
conditioning, and cooling systems. The Microsoft HoloLens™ is a semi-transparent
Head Mounted Display (HMD) that enables the display of three-dimensional holograms
in the user's field of vision based on the reconstruction of the user's real environment
[19]. Interaction between the AR hardware and the user bases on the viewing direction
(gaze) and hand gestures or voice commands. The user sees a cursor, which he or she
controls by head movements, in the center of the field of view that enables the selection
of virtual objects by performing a hand gesture, named "air tap." The "air tap" is a hand
gesture comparable to the left mouse click and is performed in three steps (see Fig. 1).
Through the first two hand movements, the user selects and holds the object
(“tap&hold”), and the user can then move it to any position by moving the hand (“tap
&move”). By lifting the index finger (gesture 3.), the user rereleases the object. The
“air tap” and the resulting hand gestures "tap&hold" and "tap&move" provide the basis
for user interaction with virtual objects when using Microsoft Hololens™. At this time,
adding individual gestures is not possible when using this hardware.
Fig. 1. Execution of the “air tap” hand gesture with the Microsoft HoloLens™ [20]
4.1 Test setup
To investigate the learning effects of using 3D gesture control, we tested five central
software functions. After the users set the starting position and the resulting superim-
position of the virtual revision plan on the real environment, they used the following
manipulation functions: adjust the height of the map display, move an object, duplicate
Moritz Quandt, David Hippert, Thies Beinke, and Michael Freitag
an object, delete an object. A moderator accompanied the test and explained all tasks
before the users were carrying them out to ensure comparability. Before performing the
tests of the five software functions, all test participants went through a tutorial on how
to perform the required hand gestures. After completing the five tasks, all test partici-
pants performed these tasks again in the same order. We recorded log files of each user
test, including the time required and the number of interactions for the execution of
individual tasks as performance measures to evaluate hand gesture-based control's
learning effects.
4.2 Task descriptions
The first task, "setting the start position", aims to ensure the accuracy of the superim-
position of the virtual building plan with the real working environment. In this step, the
user specifies the position in the room and the current viewing direction. To do this, the
user performs a "tap&hold" hand gesture after determining his or her position on the
room floor plan. A dot appears immediately at the indicated position. The user indicates
the viewing direction by pulling out an arrow in the corresponding direction
(“tap&move”). By ending the gesture, the user sets the arrow, and herewith confirms
the start position or performs the steps to set the start position again (see Fig. 2, left).
The second task, "adjust map height", contains the alignment of the displayed revi-
sion plan to the desired height in space, as shown in Fig. 2 on the right. In this case, the
task consists of moving the virtual revision plan to the ceiling. To do this, the user
performs a "tap&move" hand gesture to grab the virtual map and move it upwards. The
user can repeat this hand gesture as often as required to reach the desired height.
Task 1: Setting start position Task 2: Adjust map height
User interface Interaction User interface Interaction
Cancel
Cancel Confirm
Fig. 2. User interface and interaction pattern for tasks “setting the start position” and “adjust map
height” [21]
User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control
The following tasks serve to update the revision plan in the working environment.
This way, the service technicians learn how to use virtual revision plans and improve
the data basis for subsequent maintenance tasks (see Fig. 3).
To complete the third task, "move object", the user moves a selected object of the
revision plan from its original position to another position marked with a cross by using
a "tap&move" hand gesture. The object can be selected and moved by the user as often
as required. If the user moves the object successfully to the target position, the displayed
cross disappears, and the user has completed the task.
The fourth task, "duplicate object", is structured as follows: The user marks the ob-
ject as duplicated by performing a "tap&hold" hand gesture. This way, the user copies
the object and moves it to the target position marked with a cross by performing a
"tap&move" gesture. In this case, the differences to the task "move object" is not realted
to the execution of the hand gestures, but in the representation in the virtual revision
plan. After copying the object, the user can select the “move object” mode to adjust the
position of the duplicated object as requested.
To fulfill the fifth task, "delete an object," the user marks an object of the virtual
revision plan. By executing an "air tap," this object is marked and deleted after the
user's confirmation.
Task 3: Move object Task 4: Duplicate object Task 5: Delete object
User interface Interaction User interface Interaction User interface Interaction
Fig. 3. User interface and interaction pattern for tasks “move object”, “duplicate object”, and
“delete object” [21]
4.3 Composition of the user group
For the user study, we have recruited ten participants, all male (seven students, three
academics). The participants were in the age groups 20-25 (three participants), 26-30
(six participants), and 31-35 (one participant). All participants rated their previous ex-
perience with computers as high (one participant) or very high (nine participants). The
test users rated their previous experience with AR solutions as non-existent (5 partici-
pants), first experience (3 participants), or multiple uses (2 participants). In this case,
Moritz Quandt, David Hippert, Thies Beinke, and Michael Freitag
the previous experience of the users had no statistically verifiable effect on the results
of the user study, probably due to the sample size. In the maintenance of heating, air
conditioning, and cooling systems, the participants estimated their previous experience
as non-existent (seven participants), basic knowledge (two participants), and an inter-
mediate level of experience (one participant). Seven participants did not use visual aids;
three participants used glasses. Visual aids had no further influence on the test users
due to the insertion possibilities of the AR-glasses used. All participants were right-
handed.
5 Results and discussion
The results of the user study are shown in Fig. 4 and Fig. 5. Fig. 4 shows the average
processing times of the respective tasks for the two test runs. Fig. 5 shows the average
user interactions required to complete the task for the two test runs. All test users suc-
cessfully completed the five tasks in both test runs. This was the prerequisite for us to
ensure the comparability of the results.
For the first task, "setting the start position", we can determine that the mean pro-
cessing time to complete the task decreases slightly from the first to the second test run
(65 to 48 seconds). However, since the number of required interactions does not con-
siderably reduce, we can observe no learning effects in the use of handheld gesture
control in this task. The user's minimum number of three interactions to complete the
task is achieved by three users in the first attempt and by seven test users in the second
attempt. However, this is in contrast to the very high time and interaction requirements
of individual users. The users have to repeat the positioning several times when choos-
ing an inaccurate starting position. Repeated positioning explains the high standard de-
viation in the processing of this task by the users. From these test results, we conclude
that the interaction between users and the developed AR application is not implemented
intuitively enough at this point.
User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control
300
250
200
150
100
50
0
Task 1 - Task 1 - Task 2 - Task 2 - Task 3 - Task 3 - Task 4 - Task 4 - Task 5 – Task 5 –
run 1 run 2 run 1 run 2 run 1 run 2 run 1 run 2 run 1 run 2
mean 65,16 48,42 80,19 51,56 142,50 42,42 139,29 37,35 24,36 21,24
median 59,29 27,42 80,98 44,34 116,50 21,51 118,63 27,76 18,89 12,59
standard
23,98 49,63 34,86 27,02 82,09 52,18 71,75 24,84 16,37 30,69
deviation
Fig. 4. Box-plot diagrams of processing times for all performed tasks and the two test runs in
seconds
Users completed the second task, "adjust map height", faster and with a lower num-
ber of interactions compared to the two test runs (see Fig. 4 and 5). Due to the signifi-
cant reduction of the task processing time by an average of about 35 seconds and a
reduction of the average number of interactions by approximately two, we can observe
an apparent learning effect in hand gesture control in this task. Depending on the accu-
racy of the recorded room model, the users had to move the revision plan by about two
meters from the starting position to the ceiling. This movement required an average of
10.2 (1st test run) or 8.3 (second test run) interactions. Despite the improved perfor-
mance measures, we experience the number of "tap&move" hand gestures performed
as high for the execution of this task. Therefore, we plan to adjust the movement pa-
rameters to allow larger movements of the revision plan along the vertical axis with a
gesture's execution. With this adjustment, we expect a further improvement in the per-
formance metrics in the execution of this task.
Moritz Quandt, David Hippert, Thies Beinke, and Michael Freitag
20
15
10
5
Task 1 - Task 1 - Task 2 - Task 2 - Task 3 - Task 3 - Task 4 - Task 4 - Task 5 – Task 5 –
run 1 run 2 run 1 run 2 run 1 run 2 run 1 run 2 run 1 run 2
mean 4,90 5,80 10,20 8,30 8,60 4,70 7,20 4,10 3,20 3,00
median 4,50 3,00 10,20 9,00 7,00 3,00 8,00 3,00 3,00 3,00
standard
2,02 5,75 3,12 2,58 5,46 4,45 3,19 2,69 0,42 0
deviation
Fig. 5. Box-plot diagrams of number of interactions needed for all performed tasks
In the context of the manipulation tasks "move object", "duplicate object", and "de-
lete an object", we could observe apparent learning effects among the users between
the two test runs. Figures 4 and 5 show the corresponding execution times and interac-
tion needs of the users to fulfill the respective tasks. The users were able to reduce the
processing time for task 3 by about 70% while reducing the required interactions by
about 50%. When looking at the median, this impression manifests as, in the second
test run, only one user needed an above-average amount of time to complete the task.
In comparing the two test runs, all test users improved both in the time required to
complete the task and in the number of required interactions. Due to the analogy of the
execution of task 4 compared to task 3, we can observe similar effects in the results.
Accordingly, for the execution of task 4, we recorded shorter processing times for all
test participants in the second test run. Only two participants needed the same number
of interactions in the second run as in the first run; all other participants needed fewer
interactions in the second run. The fifth task, "delete an object," does not require any
object movement. This task could be performed by almost all participants with the min-
imum number of interactions, especially in the second test run.
With a critical look at the results of our user study, we are aware that the recorded
performance measures do not exclusively reflect the learning effects in dealing with
hand gesture-based control. The better understanding of the user's about the tasks and
the accuracy of the superposition of the virtual objects with the real working environ-
ment influenced the processing time and the number of interactions. Further, the size
and composition of the test group can be improved. A higher number of test users and
User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control
the participation of end-users from the real work environment would have led to more
founded and reliable results. Furthermore, for future user studies, the order of tasks
could be randomly selected. In this study, the order based on the workflow of the ser-
vice technicians. In connection with a larger user group, we could have eliminated that
the learning effects in using gesture control influenced the processing of individual
tasks. The last point to mention is the limited number of hand gestures, which was de-
termined by the selected hardware. The use of other hardware offers different interac-
tion possibilities. Therefore, the results of our study are not necessarily valid across
different AR hardware. Nevertheless, we see a clear added value in conducting user
studies connected with the development of mobile AR applications for industrial use.
This way, essential insights for the design of usable systems can be gained, promoting
the acceptance of the developed solutions and thus helping exploit the potentials of AR
technology.
6 Conclusion and Outlook
In this paper, we have conducted a user study to investigate the learning effects of using
a 3D hand gesture-based control system using the example of an AR application for
location-based support of service technicians in the maintenance of heating, air condi-
tioning, and cooling systems. Since there are only few guidelines for designing such
human-machine interfaces available, a user-centered evaluation can identify suitable
interaction patterns. The five introduced tasks investigated differed in the complexity
of the hand gestures to be performed and in the direct relation to 3D virtual objects in
space. The user study results confirm that the users learn direct manipulation of virtual
objects quickly since the movement of objects with the hands seems intuitive for them.
When performing multi-step hand gestures without direct relation to the real environ-
ment, we could not detect any learning effects connected with the chosen interaction
concept. We believe that multimodal interaction concepts can contribute to a more ef-
ficient performance of tasks without an object reference. The testing of such interaction
concepts represents a further research requirement for us. Besides, the present study
included a limited number of possible hand gestures, which is a result of the hardware
selection. By conducting further user studies with a hardware-independent selection of
hand gestures, we can transfer the results into general design recommendations for mo-
bile location-based AR assistance systems in the future.
Acknowledgements
The authors would like to thank the German Federal Ministry of Economic Affairs and
Energy (BMWi) for their support within the project "Bauen 4.0 – KlimAR – Aug-
mented Reality-based assistance system for the maintenance of complex heating, air
conditioning and cooling technology" (grant number 16KN062830).
Moritz Quandt, David Hippert, Thies Beinke, and Michael Freitag
References
1. Milgram, P. and Kishino, F. A taxonomy of mixed reality visual displays. IEICE Transac-
tions on Information Systems, E77-D(12), (1994).
2. Azuma, R. T. A Survey of Augmented Reality. Presence: Teleoperators and Virtual Envi-
ronments 6(4), 355–385 (1997).
3. Wang, X., Ong, S. K., and Nee, A. Y. C. A comprehensive survey of augmented reality
assembly research. Adv. Manuf. 4(1), 1–22 (2016).
4. Kipper, G. and Rampolla, J. Augmented reality. An emerging technologies guide to AR.
Syngress, Waltham (2013).
5. Fite-Georgel, P. Is there a reality in Industrial Augmented Reality? In 10th IEEE Interna-
tional Symposium on Mixed and Augmented Reality. IEEE, 201–210 (2011).
6. Palmarini, R., Erkoyuncu, J. A., Roy, R., and Torabmostaedi, H. A systematic review of
augmented reality applications in maintenance. Robotics and Computer-Integrated Manu-
facturing 49, 215–228 (2018).
7. Butz, A. and Krüger, A. (2017. Mensch-Maschine-Interaktion. Walter de Gruyter, Berlin.
8. Dünser, A. and Billinghurst, M. Evaluating Augmented Reality Systems. In: Furht, B. (ed.)
Handbook of augmented reality, pp. 289-307. Springer, New York, (2011).
9. Quandt, M., Knoke, B., Gorldt, C., Freitag, M., and Thoben, K.-D. General Requirements
for Industrial Augmented Reality Applications. Procedia CIRP. 51st CIRP Conference on
Manufacturing Systems, 1130-1135 (2018).
10. Quandt, M., Beinke, T., and Freitag, M. User-Centered Evaluation of an Augmented Reality-
based Assistance System for Maintenance. In: Procedia CIRP. 53rd CIRP Conference on
Manufacturing Systems, In Print (2020).
11. Bottani, E. and Vignali, G. Augmented reality technology in the manufacturing industry: A
review of the last decade. IISE Transactions 51(3), 284–310 (2019).
12. Lamberti, F., Manuri, F., Sanna, A., Paravati, G., Pezzolla, P., and Montuschi, P. Challenges,
Opportunities, and Future Trends of Emerging Techniques for Augmented Reality-Based
Maintenance. IEEE Trans. Emerg. Topics Comput. 2(4), 411–421 (2015).
13. Oliveira, A. C.M. and Araujo, Regina B., Jardine, Andrew K. S. A Human Centered View
on E-Maintenance. Chemical Engineering Transactions 33, 385–390 (2013).
14. Webel, S., Bockholt, U., Engelke, T., Gavish, N., Olbrich, M., and Preusche, C. An aug-
mented reality training platform for assembly and maintenance skills. Robotics and Auton-
omous Systems 61(4), 398–403 (2013).
15. Tan, Q., Chang William, and Kinshuk. Location-Based Augmented Reality for Mobile
Learning: Algorithm, System, and Implementation. The Electronic Journal of e-Learning
13, 138–148 (2015).
16. Bae, H., Walker, M., White, J., Pan, Y., Sun, Y., and Golparvar-Fard, M. Fast and scalable
structure-from-motion based localization for high-precision mobile augmented reality sys-
tems. mUX J Mob User Exp 5(1), 253 (2016).
17. Shanthakumar, V. A., Peng, C., Hansberger, J., Cao, L., Meacham, S., and Blakely, V. De-
sign and evaluation of a hand gesture recognition approach for real-time interactions. Mul-
timed Tools Appl 79(25-26), 17707–17730 (2020).
18. Ohn-Bar, E. and Trivedi, M. M. Hand Gesture Recognition in Real Time for Automotive
Interfaces: A Multimodal Vision-Based Approach and Evaluations. IEEE Trans. Intell.
Transport. Syst. 15,6, 2368–2377 (2014).
19. Liu, F. and Seipel, S. Precision study on augmented reality-based visual guidance for facility
management tasks. Automation in Construction 90, 79–90 (2018).
User-Centered Evaluation of the Learning Effects in the Use of a 3D Gesture Control
20. Microsoft Coporation, https://docs.microsoft.com/de-de/windows/mixed-reality/gestures,
last accessed 2020/07/01.
21. Stern, H., Quandt, M., kleine Kamphake, J., Beinke, T., Freitag, M. User Interface Design
für Augmented Reality-basierte Assistenzsysteme - Konzept und Anwendungsbeispiele für
eine menschorientierte Gestaltung der Nutzer*innenschnittstellen. In: Freitag, M. (ed.):
Mensch-Technik-Interaktion in der digitalisierten Arbeitswelt. Schriftenreihe der Wissen-
schaftlichen Gesellschaft für Arbeits- und Betriebsorganisation. GITO Berlin, In Print
(2020).