Page 74 - Kaleidoscope Academic Conference Proceedings 2022

2022 ITU Kaleidoscope Academic Conference

application with Unity and Vuforia [7]. Hameed et al. orientation of the camera, the target position is recognized.
introduce the development with Vuforia object targets and Targets can be e.g. horizontal surfaces, artificial markers or
tested the tracking quality in different circumstances [8]. existing objects in the environment, such as 2D surfaces or
3D models. The latter is also referred to as model-based
These approaches can be used as guidelines for developing tracking and is based on edges, textures and optical flow.
a model-based AR application with Unity and Vuforia,
but are progressively outdated with technology upgrades. In the following, particular attention will be paid to visual,
Furthermore, these approaches usually refer to only one model-based tracking with mobile devices. After target
model and are often neither practical nor realistic since detection, the virtual content is incorporated into reality in a
industrial use cases often have to consider several models correct perspective by an optical overlay of the video stream.
[9]. To the best of our knowledge, while most of the existing Virtual content can be e.g. 2D images or video, 2D or 3D
literature refer to one product, there is not yet any scientific text, 3D models as well as sound or others. As an output, the
approach proposed for the design and implementation for user can see virtual elements in correspondence to the real
multi-variant products. environment.
This research gap raises the issue of how a basic model-based The development process can be described with six major
AR prototype can be developed and implemented for steps (Figure 1) which can be traversed in an iterative process.
products with numerous variants. Subsequently, a basic understanding of the individual steps
should be conveyed. Best practices and implementation are
3. METHODOLOGY described in Section 4.2.

The Design Science Research (DSR) method according to
Oesterle et al. [10] was used as the methodology for the
development of the proposed framework and its use case.
Therefore, in an iterative process of analysis, design,
implementation and evaluation, three artifacts were created.
To simplify the development, at first a simple 3D model was
used instead of the actual product. Therefore, the first and
second artifact are abstractly illustrated by a 3D printed train
model in toy size. Only the third artifact was developed for
the real car.

The first artifact displays virtual information about a
locomotive and is a best practice for developing a basic AR
application with Unity and Vuforia. The second artifact deals
with the complexity of variant and is also illustrated by a
locomotive but with three additional wagons which represent
the different variants. The last artifact deals with the real car
and combines the findings from the first two artifacts. The
application shows the AR information about multiple car
variants. Each artifact was evaluated. In this paper, only the
performance evaluation of the second variation artifact and Figure 1 – Overview of the development process
the usability evaluation of the car application is examined of an AR application
more closely.
Step 1: Firstly, the functional and non-functional
4. DEVELOPMENT OF AN AR APPLICATION requirements towards the application must be collected.

This section presents the development process of an Step 2: Selecting the right development environment is
AR application which includes the design phase and the decisive for a successful implementation and can save a
implementation phase. lot of resources. ARCore (Google) and ARKit (Apple)
are well known but both are platform specific. There are also
4.1 Design other platforms to be considered such as Wikitude, 8 th
Wall or Vuforia. Berger et al. examined several AR tools
To be able to expand the physical world, it must first be and compared the target tracking abilities [12]. Saar
captured. This usually takes place via video stream which is analyzed six frameworks regarding tracking and rendering
recorded by a calibrated camera (visual tracking). Virtual together with Unity [13]. However, it is hard to find valid
objects can only be correctly integrated into reality if the AR comparison literature because the technology is frequently
system knows where the virtual object should appear and further developed. In addition, the choice of the AR tool
what spatial relationship the viewer (or camera) has to this depends very much on the use case. Since many AR
position [11]. After determining the position, location and tools only provide the library for tracking and rendering,

– 28 –