Visualizing Magnetic Fields in Augmented Reality

(1)

Bachelor Informatica

Visualizing Magnetic Fields in

Augmented Reality

Marcus van Bergen

June 8, 2018

Supervisor(s): dr. R.G. Belleman

Inf

orma

tica

—

Universiteit

v

an

Ams

terd

am

(2)

(3)

Abstract

Augmented reality shows promising potential in the educational scene. This thesis aims to further the research on augmented reality use in educational learning. Specifically, it researches the application of augmented reality in subjects which can be experienced as spatially complex. Therefore, we have designed an augmented reality application which can visualize magnetic fields. This application can be tested against the current methods of visualizing magnetic fields; to clarify if such an application can help lower the spatial complexity of certain subjects.

(4)

(5)

Introduction

From literature it is known that student engagement can positively benefit student learning and academic performance [4]. “Many measures of student engagement were linked positively with such desirable learning outcomes as critical thinking and grades...” Several methods have been devised to increase student engagement such as: investing in student support services, enabling students to work autonomously, etc [18]. One of the relatively unexplored areas of active engagement is the use of VR (virtual reality) and AR (augmented reality) in education [33]. AR/VR technologies are primarily used in the gaming industry, but recent studies have shown that its application in education can have many benefits [7][26].

This thesis we will further the aforementioned research by studying the effect an AR/VR appli-cation can have on learning spatially complex subjects.

1.1 Reality-Virtuality Continuum

The Reality-Virtuality Continuum, shown in Figure 1.1, is a scale which encompasses all the different possible types of mixed realities [20]. The left side of the scale begins with the real environment, which is the physical environment we reside in. The right side of the scale shows the virtual environment, which is a computer generated world that is segregated from the real environment. This rightmost virtual environment is more commonly referred to as virtual real-ity. VR is a three-dimensional (3D) environment which users can access with special electronic equipment [13]. VR contains little to no mix of the users real environment; essentially it’s an environment of its own.

Between the two environments exists mixed reality. Mixed reality is a scale that denotes the mix between the two outer realities (real and virtual). The level of mix between the two produces the mixed reality display presented.

(8)

Figure 1.1: Reality-Virtuality Continuum [20].

AR stems from the Computer Graphics field and is a method which augments the real world with computer generated objects. One advantage of AR over other mixed realities, including VR, is that it allows the user to see both the physical environment and virtual objects at once [25]. AR has actually existed for quite some time [5]. For example, heads-up-displays used by (fighter) pilots integrate flight data into their line of sight. This example not only shows the age of AR but also how AR can be used to mix computer generated data into a real environment. In the case of pilots, it leads to the users having an enriched awareness of their surroundings while not having to look at their flight instrument systems. Both this enriched awareness of surroundings and the heightened level of reality AR provides, could prove to be very useful in an educational context.

1.2 Test case selection

In order to study the effect an AR/VR application can have on learning spatially complex subjects, we must have a test case. This section will describe the requirements needed for a test case. It will then explain how magnetism satisfies these requirements, making it the chosen test case for our research.

1.2.1 Spatial Component

“Spatial ability is the capacity to understand and remember the spatial relations among objects or space.” [29] A study has shown that during geometrical spatial exercises, students with height-ened spatial abilities outperform students who are spatially inept [23]. In this study, the students were being tested on anatomy; a subject which has a spatial component. Based on this literature the first requirement for our test case is a spatial component.

Literature suggest that magnetism has a spatial component [11]. Magnetism is the name of the phenomenon observed when magnets enter each others vicinity and act upon each other. The forces these magnets exert on one another are described in 3D spaces. People with a lowered spatial ability might have difficulties understanding these described forces due to their spatial complexity. It is common practice to display these magnetic forces using magnetic field lines or magnetized iron shards (see Figure 1.2).

(9)

(a) Magnetic field line diagram. (b) Magnetized iron shards.

Figure 1.2: Commonly used (2D) magnetic field visualization techniques.

However, these methods are attempting to explain a 3D concept (the magnetic forces) using a two-dimensional (2D) space. Having to understand a 3D concept in 2D might only enhance the spatial complexity of this subject. There do exist other techniques to visualize these magnetic forces in 3D. One method to accomplish this is to submerse a magnet in a tank of oil filled with irons shards (see Figure 1.3). However, a construct like this is cumbersome, limits the interaction for the user (disallows the user to change the orientation of the magnet), and fails to display certain magnetic properties (e.g.: attraction/repulsion with other magnets).

Figure 1.3: Visualizing magnetic fields in 3D using dipole submersion [28].

1.2.2 Implementable in AR/VR

Our second requirement for our test case is that must be implementable in AR/VR. This means we must be able to simulate the test case using a computer. The output of this simulation must be visualizable in AR/VR. In the case of magnetism it’s possible to simulate the magnetic flux density (strength of magnetic field) using mathematics [11]. Move over, it’s even possible to re-write these expressions such that they return the magnetic force at a certain point in space (vector form). This suggests that magnetic forces are visualizable in AR/VR.

(10)

It clarifies that magnetism has both a spatial component, and can be implemented in AR/VR. We have therefore chosen magnetism as the test case for this research.

1.3 AR in education

In order to formulate our research question we must know which mixed reality display we will be using. The choice of mixed reality display should be based on that which yields the most benefit to education and learning. There have been several useful studies regarding AR/VR in education.

One study suggests that in the near future it is expected that AR applications will see widespread use in education [33]. This is due to the fact that AR can give learners instantaneous access to location-specific information derived from numerous sources. This idea of portable on-demand learning is very convenient in an educational context.

Another study discussed the concept of an interactive physics book. In this study the participants were divided into two groups. One group had a book which was interactive; meaning the diagrams in the books could be augmented. The other group was offered the same book but without the augmented diagrams [7]. The interactive books were aimed at making an engaging learning experience. The AR applications found in these books were built with the following qualities: “user interaction, model and texture animations, and an enhanced marker design suitable for educational books”. The results showed that the group which used the interactive book scored approximately 20% higher than those who did not. When the same subjects were administered a retention test one month after this initial experiment, those who used the interactive book continued to score 20% higher than those who did not. This study showed that participants who used AR to learn excelled in understanding the spatial complexity of the subjects. It also proved that using AR in education helps with retaining the information learned after a one month time span.

Furthermore, smart phones provide a relatively cheap and popular AR platform [26]. Studies have shown that 73% of teens have access to a smart phone [16]. These smart phones come equipped with an array of sensors (cameras, screens, CPUs, etc), which can be used together to serve an AR platform. Moreover studies also suggest that 58% of the world population will own a smart phone by 2022 [32]. Building an AR application compatible with the smart phone market can mean providing a resource, to both these populations, which can aid learning. The above literature lists several reasons why AR is suitable in an educational context. First, the portability of AR can be advantageous for students. Second, AR provides a both enriched awareness of surroundings and a heightened level of reality. Third, combining both AR and subject-matter can yield to longer learning retention and lowered spatial complexity of subjects. Finally, the accessibility smart phones grant, suggests that one AR application can see widespread use, potentially helping millions of people. With this information, we have chosen AR as the mixed reality display for this research.

1.4 Research question

This thesis studies the effect of an AR application in an educational context. We will design an AR application to aid with the spatial complexities found in magnetism. The designed solution and the sub-optimal methods of visualizing the magnetic fields will then be used to answer the following research question question: “Can an augmented reality application aid in the understanding of the properties of magnetism?”

Furthermore, to give answer on this question we will describe what steps will be taken to im-plement these magnetic properties in AR. The constraints bound to an AR application will play a large role in what methods we have chosen; this too will be described. In addition, we will

(11)

design an experiment proposal that can answer our research question. Closing this thesis will consist of concluding remarks regarding our designed application, and suggestions for future research.

(12)

(13)

CHAPTER 2

Design

The previous chapter explains why we are building an AR application that displays magnetism. This chapter will describe the design of this application. It will then list the requirements for this application. Afterwards, an extensive literature review on magnetism will be supplied. The concepts discussed in this review range from basic magnetism to expressing magnetic forces as vectors. The theory discussed in this chapter will then be verified in the form of a vector field. We will then discuss if vector fields are intuitive enough to lower spatial complexity. At the end of this chapter we will describe a method which allows multiple magnets act upon each other.

2.1 Application requirements

In order to simulate and visualize magnets and their forces our application will need to fulfill the following requirements:

1. A way to calculate magnetic forces in discrete spaces. 2. A technique to visualize coherent magnetic field lines.

3. A method which allows multiple magnets to act upon each other.

The remaining sections in this chapter will explain the concepts used to achieve these require-ments.

2.2 Basic magnetism

Before clarifying how to calculate magnetic forces in discrete spaces, we will first give a short explanation on magnets and their behaviour. Due to the scope of this thesis, we will be constrain-ing our definition of magnets to that of permanent magnets. Permanent magnets are magnetized objects which posses their own magnetic fields, even in absence of other magnetic objects [11]. This means that even when a permanent magnet is isolated it will continue to have its own mag-netic field. This magmag-netic field arises from the linear arrangement of the magnet’s molecules. Each conventional magnet (found in the physical world, not theoretical) is constructed in dipole form. Magnetic dipoles are magnets which are comprised of two poles: a north and a south pole.

Theories regarding magnetic monopoles exist however, at the moment of this writing, no such monopoles have been found in the physical world [10]. Thus, these too will be excluded from the scope of this thesis.

(14)

There are certain laws which describe the behaviour of magnets [17] [21]:

1. Given two unlike poles, north and south, the attraction trajectory extends from north to south.

2. Two identical poles, be it north or south, have a repulsive trajectory. In other words, instead of attracting they repulse away from each other.

3. Fissuring a magnetic dipole results in two smaller magnetic dipoles.

4. The larger the distance of separation between two magnets, the weaker their magnetic force is on each other; and vice versa.

These laws do not describe the full behaviour of magnets. However, for our AR application these laws will be sufficient to validate our implementation.

2.3 Magnetic forces in discrete spaces

Given the above information regarding magnets and their behaviour, the next step is to under-stand how one can express magnetic forces in discrete spaces. Given that we are working with AR, we are bound to an implementation in 3D. Doing so means that we will need a method to express these magnetic forces in a 3D space.

In physics, one commonly refers to magnetic fields as B-fields. B-fields can be expressed in various manners. We are particularly interested in an solution where our B-field is represented as a vector field. Each vector in this field will represent the magnetic force (of the magnet) at a certain point in space.

An expression which describes the magnetic field of a dipole is given as follows [11]:

Bdip(m, r) = µ0 4π 3(m · ˆr)ˆr − m r3 +2µ0 3 mδ 3_(r)

This expression does not yield the force as a vector. Therefore, we will be rewriting and simpli-fying this expression (we will remove the right side δ part of the expression), such that it does meet this criteria [19]. Firstly, we start with simplifying this expression to:

Bdip(m, r) = µ0 4π 3(m · ˆr)ˆr − m r3

This B-field expression expects m to point along the long axis of a magnet. By taking m to point along the z-axis, the expression can be re-written to:

Bdip(m, r) = µ0m 4πr3(3z ˆr/r − ˆz) = µ0m 4πr5(3zr − ˆzr 2₎

Now, the expression can be re-written to take Cartesian coordinates as inputs with the following step:

Bdip(µ0, m, x, y, z) =

µ0m

4πpx2_{+ y}2_{+ z}25

(3zx, 3zy, 2z2− x2_{− y}2₎

The magnetic moment m can expressed as m = BrV /µ0 [11]. Substituting this in our B-field

expression yields the final result [19]:

Bdip(BrV, x, y, z) =

BrV

4πpx2_{+ y}2_{+ z}25

(15)

Below an explanation of the components found in this expression:

• BrV is the Remanence field, independent on magnet geography [12]. Our formula requires

this value to be in Teslas T [19]. • x is the x axis value.

• y is the y axis value. • z is the z axis value.

• Bdipwill be a 1x3 force vector in form:

  u v w  

This B-field expression has constraints, assumptions and describes only one context. The context is that of an infinitesimally small magnetic dipole at the origin (0, 0, 0) [11]. This means we are calculating something similar to the magnetic field of a single magnetic molecule. One constraint is that our dipole is centered at the origin. There is no way to shift this dipole off-axis using only the B-field expression. However, we can use linear translations to be able to move the dipole and its field to an arbitrary point in space.

2.4 Visualizing coherent magnetic field lines

In this section we will describe two different methods for visualizing magnetic forces. We will show which method of the two is more coherent. The most coherent technique will be selected to render our magnetic field lines.

2.4.1 Vector fields

Using the aforementioned B-field expression we now have an approach to calculate the magnetic force, at any point in space, around our magnetic dipole. We will use this expression to visualize the magnetic forces in vector field form. “A vector field is an assignment of a vector to each point in a subset of space” [9]. The vectors, which will be placed at these different subsets in space, will then represent the magnetic force at that point. One issue which is immediately apparent is that the vectors closer to the origin are much larger than those further from the origin. This is logical once one understands that magnetic dipoles follow the inverse cube law [2]. The inverse cube law states that total magnetic force (F ) reduces at an inverse cube rate with each distance unit (r), or mathematically [2]:

F ≈ 1 r3

All of these varying vector lengths will lead to our displayed vector field becoming unintelligible, which adds spatial complexity to our output. One solution to this problem is to normalize all of the vectors. Each vector gets the same length, 1, but reattains their original direction. By shading each vector a color based on their previous length we can demonstrate the intensity calculated at each point in this space. Figure 2.1 displays a plot of our B-field expression using this normalization and shading solution.

(16)

Figure 2.1: Plot of magnetic fields with step sizes: 0.5, 0.4, 0.3.

With the above plot we have validated our B-Field expression. However, it’s apparent that displaying these magnetic forces in this vector field form does not help lower spatial complexity. In fact, it is quite difficult to understand in which direction the magnetic forces are traveling. This type of plot, namely a 3D arrow plot, has been described as: “ambiguous, cluttering and having a poor spatial effect. As a result, in general arrow plots are not suitable for 3D vector field visualization” [8].

To be able to visualize these magnetic forces in a intuitive manner we have chosen to implement what the paper by Post et al. describes as particle paths.

2.4.2 Particle paths

Particle paths can provide a reasonable means for visualization granted that the number of paths displayed are few [8].“If we consider a vector field in space and we have a rectilinear volume whose vertices are sampled from this vector field, one method of visualizing the field is to place particles in the field and animate their motion through the field” [15]. In order to generate these particle paths we will implement a particle advection algorithm.

Say we to take a point pi in this vector field, where pi is in voxel V . V has eight edges, at

which each edge, we can compute the magnetic force using our B-field expression. We can then use trilinear interpolation to calculate the total force vector, ~vi which is found at pi [3]. Now

that we have the force found at pi, we can use a range of methods to calculate the next point

pi+1 where our particle will move to. In our case we have chosen to implement Euler’s method

(see subsection 3.2.1 for details). Euler’s method, pi+1 = pi+ ∆t~vi, states we can find pi+1 by

(17)

Figure 2.2: Output of magnetic field lines using Euler’s method.

Figure 2.2 shows the visualization of our magnetic dipole described in section 2.3. This visual-ization uses the B-field expression to generate the magnetic forces in vector field form. We then run the particle advection algorithm on this vector field, which yields the four particle paths seen in the figure. These particle paths produce a coherent representation of the magnetic field lines as desired. We will be using this visualization technique in our final implementation.

2.5 Multiple magnets: the superposition principle

The final design requirement for our application, is that it supports multiple magnets acting upon each other. One property of magnetic fields is that they follow the superposition principle [11].“The superposition principle, states that, for all linear systems, the net response caused by two or more stimuli is the sum of the responses that would have been caused by each stimulus individually” [31].

Pertaining to vectors, the superposition principle states that vectors respect the following prop-erties:

Additivity : F (x1+ x2) = F (x1) + F (x2)

Homogeneity : F (ax) = aF (x), for scalar a

In our case, the additivity property can be used to calculate the vectors which represent the forces of two (or more) magnets acting upon each other. Say we were given two magnets, with two (overlapping) vector fields. If we were to iterate through every point in this overlapping area, we would have two force vectors, from two magnets, at each point. By making use of the

(18)

additivity property, we can add the two vectors together F (x1) + F (x2) which produces a third

vector F (x1+ x2). This third vector, better known as the net vector, represents the force of

the two magnets acting upon each other. We expect this net vector to reproduce the behaviour described by the behavioural laws of magnets (see section 2.2 for details).

(19)

CHAPTER 3

Implementation

In the previous chapter we describe the methods required to build our AR application for mag-netism. This chapters starts by listing the constraints bound to our AR application. We will then explain the ways we have implemented each method, from the previous chapter, to best meet these constraints. Furthermore it will provide an explanation of the SDKs (Software devel-opment kits) used to build this application. Finally, we will provide a brief overview on how we have structured the entire application using these SDKs and our own methods.

3.1 Constraints

AR applications are best experienced when they are responsive [14]. In order to factor out responsiveness being a hindrance in our research, our application must meet certain criteria to categorize it as responsive. In terms of FPS (frames per second) the human brain begins to perceive something as motion when it is being shown at greater than 12 FPS [22]. In the year 1930, 24 FPS became the standard for film.

With this in mind, we are setting 24 FPS as the lower bound frame rate of our application. This means that while the application is running there is a maximum of 41.66 milliseconds for each frame to be produced. During these 41.66 milliseconds a whole range of actions need to take place. These span from processing the camera frames to drawing the correct pixels on the screen (see Figure 3.1 for an overview). In order to achieve 24 or higher FPS performance in our application, we are looking to cut computational time where unnecessary.

Furthermore, we have mentioned that our application will be designed for the mobile market. This means one can assume that the computing power we are working with is limited to that of modern smart phones. Due to this limitation certain programming paradigms have been excluded out of our implementation. For example: Nvida’s CUDA (Compute Unified Device Architecture), which is a framework that provides parallel computation via GPUs, will not be used due to its limited support in the mobile market [6].

3.2 Implementing swift particle advection

In subsection 2.4.2 we explained why we are using particle paths as our main method to display magnetic field lines. This section also briefly stated that we would be using Euler’s method as the approximator to calculate the next point pi+1. The choice for Euler’s method was based on

(20)

3.2.1 Particle advection: computational constraints and considerations

The paper by Joy stresses the error estimate which Euler’s method is bounded by, namely O(∆t2₎3 _{[15]. We have considered other methods such as the Improved Euler Method and the}

(fourth-order) Runge-Kutta Algorithm, which both have proven to generate more accurate out-comes. The error estimate bound to both of these methods are O(∆t3_{) and O(∆t}4_{) respectively.}

Both are considerably lower than that of Euler’s method.

Even so, as mentioned before we are seeking to minimize extra computation where possible. In our case, both the improved Euler method and the Runge-Kutta Algorithm require more computations per point approximated [15].

With respect to particle paths in our application, the following needs to happen within 41.66 milliseconds:

1. For one particle path, we need to approximate an entire trajectory of points. This can be upwards of 2000 points per path.

2. As seen in Figure 2.2, we are seeking to approximate multiple particle paths per dipole. Meaning the above statement has to happen multiple times.

3. The application needs to visualize multiple magnets. This implies that the above statements need to happen multiple times for each magnet.

For these reasons we have chosen Euler’s method, because it gives us the highest chance of stay under this threshold. However, if our application were a computational model that was not bound by processing time constraints, we would have chosen a different approximation algorithm.

3.2.2 Further optimizations

Furthermore, instead of generating all of the force vectors defined at each linearly spaced x, y, z, of complexity O(n3_{), we have chosen to only generate the force vectors which are needed during}

the trilinear interpolation. While the particle advection is running we know exactly which edge points are required for each pi. Thus, we only need to calculate each to be used force vector

once. This is another implemented optimization that has the goal of achieving a responsive AR application. Finally, we can store each force vector in a dictionary where the key is the coordinate and the value the vector. This is done in order to reduce seek time.

3.3 AR and Unity

The aforementioned methods provide us sufficient tools to make an AR implementation of mag-netic field lines (represented by particle paths). For the AR application we are using Unity and Vuforia. Unity, is a cross-platform game engine commonly used to develop video games [26]. It is possible to script in Unity by using either C# or JavaScript; in our case we will be using C#. Unity allows one to write C# classes which can then be linked to gameobjects. In addition, Unity can be expanded to support AR/MR by installing the Vuforia SDK. Vuforia is a software development kit which enriches Unity with tracking software, HMDs (head mounted displays), AR cameras, etc... [27]

For the first step of our implementation we will be making use of Vuforia’s marker-tracking software. This software allows us to take an arbitrary amount of images and make a marker database of them. Each marker in this database can then be linked to a gameobject. Gameob-jects, as the name implies, are objects which can represent almost anything in Unity. In our case we can create two game objects which represent two different virtual permanent magnets (see Figure 2.2). When the markers have been imported into Unity, we can use the C# scripts to link the methods from the previous sections to the gameobjects.

(21)

3.3.1 Drawing Particle Paths

The Unity library provides an object known as a lineRenderer. A lineRenderer object can be used to draw line segments between a list of given points. Knowing this, it’s possible to use the lineRenderer to draw the trajectories (particle paths) computed during the particle advection. In order to get this working we will needed to to keep a list of the different points calculated and pass this to the lineRenderer object. The lineRenderer will then draw small line segments between all the points returned from the particle advection. With this technique we can display one particle path in AR. The operation can then be repeated for several particle paths for each dipole.

3.3.2 Local - Global coordinates

Unity makes use of what is known as world coordinates. These coordinates are that of the virtual world where our gameobjects (dipoles) are located. When we generate force vectors of a dipole using this B-field expression these force vectors are by default centered around the origin. Unity’s world coordinates know only one origin, thus our vectors are placed at the wrong location. In order to place these vectors at the right location, namely around their dipole, we need to apply a linear translation to the points where the vectors take place. This translation will be the difference of (x, y, z) points between this global origin to and the local origin of the dipole. This distance can be described as (∆x, ∆y, ∆z). Mathematically, our linear translation can now be described as: (x, y, z) → (x + ∆x, y + ∆y, z + ∆z). Where (x + ∆x, y + ∆y, z + ∆z) are our new coordinates. After this operation the force vectors will now be at their correct location in the world space. One optimization we can apply is one where two game objects are not overlapping. When this is the case we can choose to only translate the particle path from the origin. However, when this is not the case we will have to apply our linear translation on the vector field coordinates. To identify when our dipoles are overlapping we will need a method to track the gameobjects in Unity. This will be explained in the next subsection.

3.3.3 Marker tracking and dipole interaction

Now that the dipoles are correctly visualized on their markers, we will describe how we check when the magnets overlap. We will start with tracking the locations of the markers in the virtual world by calling the gameObject.transform.position function. This function returns the coordinates of the gameobject in question (in the virtual world). Each dipole’s magnetic field has a bounding box, which is described by its outermost vectors. With these bounding boxes we can check when two dipoles overlap. This is accomplished by calculating whether the bounding boxes of two dipoles intersect. If this is not the case then magnetic fields are not acting upon each other.

However, should they have overlap then we know the magnets are acting on each other. We can then call a function which takes two or more dipoles and finds their overlapping area. At these coordinates we will apply the additive property described in section 2.5 to calculate the net vectors. For each vector field we will use the net vectors instead of the calculated force vectors where these exist. Seeing as the vector fields have changed the particle advection function should also be run again. The new particle paths will display the behaviour described by the magnetic behavioural laws.

3.3.4 Application overview

Figure 3.1 best summarizes our application. This figure explains the bridges between each implemented component. The pieces which are not bold (motion tracking, rendering, display) are software used from the SDKs. The rest are the implemented methods explained above.

(22)

(23)

CHAPTER 4

Designed Experiment

The previous chapter has given a thorough explanation of the implemented methods used in our AR application. In the beginning of this thesis we clarified that we would limit our definition of magnets to that of permanent magnets. We also explained that magnetic field lines are meant to be a tool to aid the user understand the behavioural laws of magnetism. Given this information, we will limit our experiment to that of people with limited to no knowledge of magnetism. This experiment will be aimed at answering our research question: “Can an augmented reality application aid in the understanding of the properties of magnetism?”. Furthermore, we will be describing the steps we took to design a testable environment; to give answer on this research question.

4.1 Method

Given our research question, we are most interested in measuring the effect that our AR applica-tion has on participants’ spacial understanding of magnetism. Our experiment will consist of two groups: one group which will learn using the AR application and another group which will be given the current methods of understanding dipoles (diagrams and iron shards, see Figure 1.2). The size for both groups will be around five to ten persons.

Both groups will be given a brief explanation on the laws which describe the behaviour of permanent magnets (control variable). However, we will provide no visual stimulation during this explanation. Rather, the visual stimuli will be our independent variable. We will grant the participants of both groups a set amount of time (around 10 minutes) to interact with their visual stimulus; either diagrams and irons shards or the AR application. During this time, we expect the participants with a lowered spatial ability to gain a better insight as to how these behavioural laws apply in the physical world. After this set time has passed, we will require the participants to partake in a small test and feedback round.

4.2 Designed Test

Our test is designed to measure how well our participants understood the magnetic properties explained to them. In order to do this we need to challenge them with scenarios that are both spatially complex and relevant to magnetic behaviour. Our test consists of eight 2D and eight 3D figures (on paper). These figures will contain magnets in multiple scenarios. Each scenario will have one figure with the correct field lines drawn and another with the incorrect field lines drawn. The participants will then have to choose the correct figures (multiple choice) by making use of their recently acquired knowledge.

(24)

Furthermore, we will also be collecting qualitative feedback from our participants. A study by Wickens concluded that if designed correctly, a VR application can greatly help users with learning [30]. It also showed that by using a VR application, students were more motivated to finish the task at hand. The leads to the students being more interested in mastering the subject that they learning. Furthermore, by optimizing the interface and making it more natural, the cognitive load of interacting with the application was reduced. This resulted in the subjects having more cognitive power to focus on learning.

Given this information we realize it is important to measure how well of an experience our users had with our application. We will be using an Intrinsic Motivation Inventory (IMI) as our measuring device for these questions.“(IMI) is a multidimensional measurement device intended to assess participants subjective experience related to a target activity in laboratory experiments” [24]. From the large supplied IMI questionnaire we have chosen select questions and modified them in order to measure this subjective experience. We also believe that it is important to gather feedback regarding the different levels of spatial complexity experienced by each participant. We can use this information in our analysis to filter our results based on the participants’ spatial complexities.

Our designed test can be found in the appendix at pages 31-35.

4.3 Analysis

After the data has been collected we will perform an analysis on our results. We will start by grading the figures part of the test. Each user will get one point for a correct answer. We will then calculate the average grade of both groups to see which group scored highest.

After which, we will create another analysis. Here, we will be using the results collected from the participants who experience magnetism as spatially complex. Again, we will be splitting the groups into two based on the visual stimuli used by the participants. The results should clarify if people with a lower spatial ability scored higher when using the AR application, or when using the classical methods.

(25)

CHAPTER 5

Conclusion and discussion

The global objective of this thesis was to understand if an AR application can be used to better learning in an educational context. In our research, we used understanding basic magnetic properties as a test case for the educational context. Doing so yielded the research question: “Can an augmented reality application aid in the understanding of the properties of magnetism?”

5.1 Conclusion

Throughout this research we have concluded that there can be many benefits to using a well de-signed AR application for learning. Firstly, AR applications allow us to visualize abstract or non perceivable concepts [1]. Besides magnetic fields, there are many concepts in the (educational) universe which are unperceivable. Visualizing these unperceivable objects, using AR, allows the users to see them as if they are actually there. Being able to see these objects could be the extra stimulus needed to lower the spatial complexity of a concept.

Furthermore, we have also realized that choosing the correct visualization technique is very important [8]. Simply visualizing something in AR is not sufficient to lower its spatial complexity. In our case, we implemented particle paths to display the magnetic field lines, of a magnets, instead of using arrow fields. These particle paths gave a better representation of how the magnetic field lines should look like. Being able to use different visualization techniques allows the designer to fine-tune their AR application, such that the information presented can be in its lowest spatially complex state. Developers should take this into consideration when implementing an AR application.

In addition to this, AR can offer the users a more interactive experience. In our case the users can use the markers to both rotate and move the virtual magnets. The classical methods, such as those in Figure 1.3, limits the users interaction to that of one degree of freedom. These extra degrees, and the heightened interaction AR offers, can be a contributing factors to users learning better.

5.2 Discussion

The methods described in our thesis do raise certain points which need to be discussed. First of all, there can be other complications to understanding a problem other than spatial complexity. This thesis assumes that spatial complexity is the obstruction to a persons understanding of a certain subject. We try to solve this obstruction by using computer visualizations in the form

(26)

of AR. However, if the obstruction is something other than spatial complexity, the role AR can play in helping a user is unknown.

Moreover, the role that user experience plays in persons successfully learning via AR must be taken in to consideration. Even though the study of Wickens concluded that participants are not discouraged by usability problems, bound to application interfaces, we question this behaviour in long term use [30].

5.3 Future research and closing remarks

Given the time constraints for this thesis certain concepts had to be excluded from our research. We believe that markerless tracking can hugely benefit an AR application such as the one de-scribed in this thesis. For future research, we suggest using real magnets as the “markers” in such an application. Doing so will result in users being able to both see and feel the magnetic forces at the same time. This adds an extra dimension to the application which might result in an even more engaging learning experience. This can possibly lead to users understanding magnetism even better.

Finally, it would be interesting to see how well participants would score given a test in AR. The test this thesis describes aims to measure how well our participants understood magnetism on a 2D surface. A test designed in AR, might provide more insight whether an AR application can be used to better learning in education.

(27)

Bibliography

[1] Sasha A Barab et al. “Virtual solar system project: Building understanding through model building”. In: Journal of research in science teaching 37.7 (2000), pp. 719–756.

[2] Xavier Borg. “The inverse cube law for dipoles”. In: (2009), 2pp.

[3] _{Paul Bourke. Interpolation methods. 1999. url: http://paulbourke.net/miscellaneous/} interpolation/.

[4] Robert M. Carini, George D. Kuh, and Stephen P. Klein. “Student Engagement and Stu-dent Learning: Testing the Linkages*”. In: Research in Higher Education 47.1 (Jan. 2006), pp. 1–32. issn: 1573-188X. doi: 10.1007/s11162-005-8150-9. url: https://doi.org/ 10.1007/s11162-005-8150-9.

[5] T. P. Caudell and D. W. Mizell. “Augmented reality: an application of heads-up display technology to manual manufacturing processes”. In: Proceedings of the Twenty-Fifth Hawaii International Conference on System Sciences. Vol. ii. Jan. 1992, 659–669 vol.2. doi: 10. 1109/HICSS.1992.183317.

[6] Daniel Eran Dilger. How AMD and Nvidia lost the mobile GPU chip business to Apple — with help from Samsung and Google. 2015. url: https://appleinsider.com/articles/ 15/01/23/how- amd- and- nvidia- lost- the- mobile- gpu- chip- business- to- apple-with-help-from-samsung-and-google-.

[7] Andreas D¨unser et al. “Creating interactive physics education books with augmented real-ity”. In: Proceedings of the 24th Australian computer-human interaction conference. ACM. 2012, pp. 107–114.

[8] Jarke J. van Wijky Frits H. Post. “Visual Representation of Vector Fields”. In: ONR Workshop on Data Visualization (1993). url: https://pdfs.semanticscholar.org/ 3ed6/76d391b81d089cbcc1e5469de73acd33c545.pdf.

[9] Antonio Galbis and Manuel Maestre. Vector analysis versus vector calculus. Springer Sci-ence & Business Media, 2012.

[10] G Giacomelli and L Patrizii. “Magnetic monopole searches”. In: Non-Accelerator Astropar-ticle Physics. World Scientific, 2005, pp. 129–142.

[11] David J Griffiths. Introduction to electrodynamics. Prentice Hall, 1999.

[12] _{Supermagnete inc. Physical Magnet Data. url: https://www.supermagnete.nl/eng/} data_table.php.

[13] Jerry Isdale. “What is virtual reality”. In: Virtual Reality Information Resources http://www. isx. com/˜ jisdale/WhatIsVr. html 4 (1998).

[14] H. S. Jang et al. “Augmented reality with high frame rate for low computational power devices”. In: 2011 IEEE International Conference on Consumer Electronics -Berlin (ICCE-Berlin). Sept. 2011, pp. 274–275. doi: 10.1109/ICCE-Berlin.2011.6031808.

[15] Kenneth I. Joy. “NUMERICAL METHODS FOR PARTICLE ADVECTION IN VECTOR FIELDS”. In: (N.A.). url: http://graphics.cs.ucdavis.edu/~joy/ecs289h/Papers/ Numerical-Methods-for-Particle-Tracing-in-Vector-Fields.pdf.

[16] Amanda Lenhart. “Teen, Social Media and Technology Overview 2015”. In: (2015). [17] _{Jim Lucas. What Is Magnetism? Magnetic Fields & Magnetic Force. 2015. url: https:}

(28)

[18] _{PhD Maryellen Weimer. 10 Ways to Promote Student Engagement. 2012. url: https :} //www.facultyfocus.com/articles/effective-teaching-strategies/10-ways-to-promote-student-engagement/.

[19] Brian Moths (https://physics.stackexchange.com/users/23785/brian moths). Calculating magnetic field. Physics Stack Exchange. URL:https://physics.stackexchange.com/q/311169 (version: 2017-02-11). eprint: https : / / physics . stackexchange . com / q / 311169. url: https://physics.stackexchange.com/q/311169.

[20] Akira Utsumi Fumio Kishino Paul Milgram Haruo Takemura. Augmented reality: a class of displays on the reality-virtuality continuum. 1995. doi: 10.1117/12.197321. url: https: //doi.org/10.1117/12.197321.

[21] _{Kate Prudchenko. The Laws of Magnets. url: https://www.techwalla.com/articles/} instructions-direct-tv-satellite-dish.

[22] Paul Read and Mark-Paul Meyer. Restoration of motion picture film. Butterworth-Heinemann, 2000.

[23] K Rochford. “Spatial learning disabilities and underachievement among university anatomy students”. In: Medical education 19.1 (1985), pp. 13–26.

[24] _{selfdeterminationtheory.org. Intrinsic Motivation Inventory (IMI). N.A. url: http : / /} selfdeterminationtheory.org/intrinsic-motivation-inventory/.

[25] Brett Shelton and Nick Hedley. “Using augmented reality for teaching Earth-Sun relation-ships to undergraduate geography students”. In: (Feb. 2002), 8 pp.

[26] _{Unity Technologies. Unity - Products. 2018. url: https://unity3d.com/unity.}

[27] _{Unity Technologies. Unity - Vuforia. 2018. url: https://docs.unity3d.com/Manual/} vuforia-sdk-overview.html.

[28] UniServeScienceVIDEO. “PH EM MF DEMO 70012A V0125 3D Magnetic Field Demon-strator Oil Filled Cube Bar Magnet Still”. In: (2013). url: https://www.youtube.com/ watch?v=V-M07N4a6-Y.

[29] John Hopkins University. “Spatial ability”. In: (N.A.), 1 pp.

[30] Christopher D Wickens. “Virtual reality and education”. In: Systems, Man and Cybernet-ics, 1992., IEEE International Conference on. IEEE. 1992, pp. 842–847.

[31] R. Wolfson. Essential University Physics. MasteringPhysics Series. Pearson Addison-Wesley, 2007. isbn: 9780805392128. url: https://books.google.nl/books?id=6bqPAQAACAAJ. [32] Y Wu. “Global smartphone user penetration forecast by 88 countries: 2007-2022”. In:

Wireless Smartphone Strategies Services (2016).

[33] Steve Chi-Yin Yuen, Gallayanee Yaoyuneyong, and Erik Johnson. “Augmented reality: An overview and five directions for AR in education”. In: Journal of Educational Technology Development and Exchange (JETDE) 4.1 (2011), p. 11.

(29)

(30)

(31)

Test on Magnetic Dipoles

NAME:

TESTNR:

Figures

Circle the correct answer for each question.

Question 1:

(a) (b)

Question 2:

(32)

Question 3:

(e) (f)

Question 4:

(33)

Question 5:

(i) (j)

Question 6:

(34)

Question 7:

(m) (n)

Question 8:

(35)

Feedback on experiment

For each of the following statements, please indicate how true it is for you, using the given scale.

Spatial Complexity

1. I find magnets and their forces spatially complex.

1 2 3 4 5 6 7

not at all somewhat

true very true

2. I felt that the visual stimulus helped lower the task’s spatial complexity.

1 2 3 4 5 6 7

not at all somewhat

true very true

3. I believe an AR application can help lower the task’s spatial complexity.

1 2 3 4 5 6 7

not at all somewhat

true very true

AR interaction

1. I felt frustrated while doing the task.

1 2 3 4 5 6 7

not at all somewhat

true very true

2. I found that the application was distracting me from accomplishing the task.

1 2 3 4 5 6 7

not at all somewhat

true very true

3. I would have preferred a HMD (head mounted display) for this task.

1 2 3 4 5 6 7

not at all somewhat

Visualizing Magnetic Fields in Augmented Reality

Bachelor Informatica