=Paper=
{{Paper
|id=Vol-2308/isee2019paper06
|storemode=property
|title=Teaching Wearable Device Development with the Wearables Development Toolkit
|pdfUrl=https://ceur-ws.org/Vol-2308/isee2019paper06.pdf
|volume=Vol-2308
|authors=Juan Haladjian,Bernd Bruegge
|dblpUrl=https://dblp.org/rec/conf/se/HaladjianB19
}}
==Teaching Wearable Device Development with the Wearables Development Toolkit==
https://ceur-ws.org/Vol-2308/isee2019paper06.pdf
Teaching Wearable Device Development with the
Wearables Development Toolkit
1st Juan Haladjian 2nd Bernd Bruegge
Chair for Applied Software Engineering Chair for Applied Software Engineering
Technical University of Munich) Technical University of Munich)
Munich, Germany Munich, Germany
haladjia@in.tum.de bruegge@in.tum.de
Abstract—This paper introduces the Wearables Development
Toolkit (WDK), a set of tools to support the development of
wearable device applications. It lowers the entrance barrier to
wearable device development. We discuss our experiences in
leveraging the WDK to teach wearable device development to
students of computer science.
Index Terms—software engineering, wearable computing,
wearables development toolkit
I. I NTRODUCTION
The growing number of students of computer science calls
for new teaching methodologies that are able to cope with
larger number of students while maintaining the teaching
quality. Digital technology has made it possible to transmit Fig. 1. Wearable sensor kit developed by InteractiveWear. Source:
http://www.interactive-wear.com/
content to large numbers of students. However, actively en-
gaging students in the learning process remains a challenge.
Courses where students actively work on a project are usually
associated with high supervision costs and therefore scale less annotation of data. After this, developers usually develop
well to larger number of students. A particular challenge for and evaluate different signal processing and machine learning
instructors is combining their teaching and research responsi- methods. Finally, the application is deployed into the actual
bilities. device. The WDK consists of four components: the Wearable
Our research focus is on wearable computing. We develop Sensors Platform, the Data Annotation Tool, the Visualization
wearable device applications, which make use of sensor data tool and the Evaluation tool.
to extract relevant information from the user or her context.
The Wearable Sensors Platform is a collection of wearable
For example, we developed a wearable device that uses a
sensors which can be plugged into a sensor hub and configured
motion sensor to detect lameness in dairy cattle [1], [2].
over an iPhone App. This enables users to collect data without
The development of wearable device applications requires
having to design or assemble a new sensor or to develop
multidisciplinary highly-specialized knowledge (e.g. electrical
a firmware that stores data. Figure 1 shows the sensor hub
engineering, computer and data science).
and different hardware components. The Data Annotation
The WDK is a collection of wearable sensors (see Figure
Tool is used to automatize the data annotation process by
1) and tools to facilitate the development process of wearable
synchronizing and displaying the sensor data together with
device applications. The toolkit is meant to guide students
reference markers enabling the user to annotate events in the
during the development process and to enable them to study
time series signal. Figure 3 shows the Data Annotation Tool.
possible design solutions while saving time in implementation
The Visualization tool enables users to understand the sensor
details. In this paper, we discuss our experience in teaching
data as well as the effects different signal processing methods
wearable application development to students of computer
have on the data, which is critical for most activity recognition
science using the WDK.
application. The Evaluation tool lets users quickly configure a
II. W EARABLES D EVELOPMENT T OOLKIT chain of computations in order to assess its performance. The
WDK is open source1 .
Figure 2 shows the different activities in the development
process of a wearable device application. Most wearable
device applications extract information from sensor data. The
development process usually starts with the collection and 1 https://github.com/avenix/WDK
ISEE 2019: 2nd Workshop on Innovative Software Engineering Education @ SE19, Stuttgart, Germany 27
Fig. 2. Wearable device development activities.
Fig. 4. Data Visualization Tool displaying the acceleration of a lacrosse
goalkeeper while performing several training exercises.
application. As students start engaging in activities for which
the WDK can spare them time, their instructor demonstrates
Fig. 3. Data Annotation Tool displaying the acceleration collected by a Inertial
Measurement Unit attached to a limb of a cow. The strides performed by the the relevant tools within the toolkit. As students use the WDK,
cow have been annotated. new requirements for the toolkit are identified, which are
usually analyzed by an instructor and implemented by other
students in the next term.
III. T EACHING M ETHOD
IV. C ONCLUSIONS
Since 2011, we have supervised over 30 Bachelor’s and
Master’s theses in computer science at the Technical Univer- The WDK enables students to reuse functionality and focus
sity of Munich. Most of these theses comprise the development on the novel aspects of their projects. The different tools
of a wearable device application or a feature of the WDK and documentation guide students through the development
itself. In this section, we describe how we leverage the number process, thus relieving instructors.
of students to contribute to our research in wearable device R EFERENCES
development.
[1] J. Haladjian, J. Haug, S. Nüske, and B. Bruegge, “A Wearable Sensor
In the beginning of each semester, we provide students System for Lameness Detection in Dairy Cattle,” Multimodal Technolo-
a tutorial on activity recognition with wearable devices2 . gies and Interaction, vol. 2, no. 2, p. 27, 2018.
[2] J. Haladjian, Z. Hodaie, S. Nüske, and B. Brügge, “Gait Anomaly
Students are usually able to finish this tutorial within a Detection in Dairy Cattle,” in Proceedings of the Fourth International
day. Afterwards, the students start working on a particular Conference on Animal-Computer Interaction, ser. ACI2017. New
York, NY, USA: ACM, 2017, pp. 8:1—-8:8. [Online]. Available:
2 Tuorial on activity recognition with wearable devices: http://doi.acm.org/10.1145/3152130.3152135
https://github.com/avenix/ARC-Tutorial/
ISEE 2019: 2nd Workshop on Innovative Software Engineering Education @ SE19, Stuttgart, Germany 28