Analysis of the Emotions Experienced by Learning Greedy Algorithms with Augmented Reality Maximiliano Paredes-Velasco Mónica Gómez Rios Angel Velázquez-Iturbide Universidad Rey Juan Carlos Universidad Politécnica Salesiana Universidad Rey Juan Carlos Móstoles, Spain Guayaquil, Ecuador Móstoles, Spain maximiliano.paredes@urjc.es mgomezr@ups.edu.ec angel.velazquez@urjc.es Abstract—Students have difficulties in understanding through teaching methodologies based on gamification algorithm subjects, in particular how the source code of [21,22] and collaborative learning [23]. The objective of this algorithms proceeds to solve problems. This article presents an work is to conduct an exploratory study of the advancement augmented reality tool intended to assist in learning greedy of knowledge and the emotions that students experience while algorithms. Students use their smartphone’s camera to focus the they study greedy algorithms using augmented reality source code of Dijkstra’s algorithm as written on paper, and the technologies. An experience was conducted in a classroom tool shows how the algorithm works. An experience was where students used an augmented reality tool on their conducted in the classroom to assess students’ emotions and smartphones, called RA-AVD (Realidad Aumentada – knowledge level. The results show that positive emotions Algoritmo Voraz de Dijkstra, in English Augmented Reality experienced by students were almost twice as intense as negative – Dijkstra’s Greedy Algorithm), along with paper notes ones. Despite the complexity of the task (i.e. understanding Dijkstra’s algorithm), the level of enjoyment of students was provided by the teacher. continuous during the experience. However, the anxiety II. METHODOLOGY experienced by students was the double than at the beginning. Keywords— Emotions, learning, Augmented reality, Greedy A. Educational objective and context Algorithm. The objective of the experience is to assess the level of knowledge and the positive and negative emotions that I. INTRODUCTION students experience by using the RA-AVD tool in solving In algorithm design subjects of university computer optimization problem. The experience is conducted in the science degrees, the development of algorithms that solve context of the Computer Science Degree at the Salesian optimization problems is usually studied. In this educational Polytechnic University of Ecuador, specifically in the area of context, learning greedy algorithms is one of the most Data Structures. At some point in this subject, students have complex topics for the student, not being able to translate into to learn and develop greedy algorithms. In particular, they source code the elements of their development (i.e., set of must understand Dijkstra’s algorithm [24], which solves a candidates, selection function, feasibility function, objective classic graph problem: determining the minimum length path function) [1]. This difficulty may affect the emotional state of from a source node to the rest of the nodes in the graph. students and cause them to become discouraged during the Students participated voluntarily and they had no previous learning process. Motivation and emotions in learning play a contact or knowledge about greedy algorithms, although they fundamental role since they influence memory and logical knew how to program and have basic knowledge of the Java reasoning and help improve attention [2]. Today, programming language. neuroscience research helps us in understanding how the brain B. Variables and measurement instruments works and the influence and importance of emotions to improve learning [3,4]. Along with this, if a student is not The variables measured were the emotions experienced by predisposed to learn, or he/she experiences strong negative the students and the level of knowledge they acquired after the emotions, it is unlikely that he/she will be able to achieve experience. The level of knowledge is measured by a his/her full potential. Furthermore, learning is characterized as knowledge test on the behavior of Dijkstra’s algorithm. The a cognitive and motivational process [5], where emotions may test was formed by 5 multiple-choice questions where each affect both the intrinsic and extrinsic motivation of the student question was scored with maximum 2 points. [6,7]. The instrument used to measure emotions was AEQ Not only emotions can influence learning outcomes. The (Achievement Emotions Questionnaire), which is a consistent use of new technologies plays a relevant role in learning [8]. and validated scale in the educational context [25]. Taking Specifically, augmented reality (AR) technology may into account the principles of neuroeducation [26], the improve learning performance [9] and constitutes a emotional variables to be measured can be classified into two technology option with great potential and effectiveness to types: 1) activation emotions, which are the emotions that activate positive emotions in students [10]. Augmented reality produce a higher degree of agitation (fear, anxiety, anger, etc.) not only provides immersive experiences of visibility or and 2) deactivation emotions, which produce lower agitation observation [11-13], but it may also contribute to student’s (depression, calm, boredom, etc.). In addition, these emotions feeling of greater satisfaction [14], improved usability [15] can be classified by the positive (pleasant sensation) or and reduction of his/her cognitive load [16] in the use of negative impact (unpleasant or uncomfortable sensation) that technological tools during the learning process. Augmented they produce on participants. Overall, up to four classes of reality technologies have been applied to programming emotions could be identified. learning at different educational levels, showing satisfactory The AEQ scale measures these emotions by offering a results: from early ages [17,18], high school and university series of statements about the participant’s emotional state and students [19] to professional adults [20], being applied mainly students must assess the degree to which they describe their Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). emotions and feelings. Each statement is rated in a Likert Figure 1). In summary, the student can read code in the sheets scale, which ranges from very little (value 1) to extremely (a static description of the solution) and can watch an (value 5). In Table I, the variables of emotions measured with animation of the algorithm on the mobile (a dynamic AEQ are detailed (the number in parentheses corresponds to demonstration of the solution). the total of items to be assessed for that emotion). Note that the test only addresses three kinds of emotions and that it consists of 75 items which measure 8 emotions. TABLE I. MEASURED VARIABLES OF EMOTIONS AND THE NUMBER OF ITEMS Activation Emotion Activation Deactivation Positive Enjoyment (10) Hope (6) Pride (6) Negative Anger (9) Hopelessness (11) Anxiety (11) Boredom (11) Embarrassment (11) Fig. 1. Extension of the source code with the visualization of the graph AEQ scale is organized so that students assess at the end detected by means of augmented reality. of the session their emotional state at three different times: The tool has a user interface that allows to control step by 1) Before starting the learning task: the student assesses step the execution of the algorithm, obtaining a display of its how he/she feels right at the beginning of the experience. execution trace, as in a debugger (see Figure 2). This trace 2) During the task: the student assesses how he/she feels screen shows the graph declared in the source code at the top while doing the learning task. and the trace table along with the step-by-step execution 3) After the Task: the student rates how he/she feels after controls below (see Figure 2). completing the experience. Figure 2 shows the trace in an intermediate state to Finally, the opinion of participants about the experience calculate the minimum paths from node 0 to the other nodes. was collected by asking some of them their opinion in a short Notice that some nodes present a label formed by brackets in interview. the form of [A,B]N, where A is the distance of the path from the source node to the node, B is the predecessor node from C. Phases that path and N is the resolution stage number. The algorithm In the first phase, students attended at a laboratory and they solves the problem in stages, in such a way that at each stage received on printed sheets the markers to present the it analyzes possible new paths between the source node and information with augmented reality. These sheets contained each remaining node, recording a new path if it is shorter than the Java source code for Dijkstra's algorithm. the current path. Therefore, these labels are dynamically generated as tracing progresses, and they are replaced every Subsequently, students downloaded the application to time the algorithm finds a shorter path. In this case, the path their mobile devices and a brief explanation was given about that partially expresses the bracket label is discarded (by the task to be carried out (for 10 minutes), which consisted of displaying a horizontal line that crosses it out) and a new label reading and interpreting the source code that appeared on the is shown by the node, representing the new path. sheets provided. An example may assist in better understanding this While reading the Java code on paper, students could use notation. In Figure 2 we can see that node 3 has the label [8,1]4 the mobile to focus parts of the code and receive assistance with a crossing line, which means: from the application. This phase lasted 30 minutes.  Number 8: length of the path from source node to Once this phase was finished, the students proceeded to node 3 (labeled node). carry out the evaluation of acquired knowledge and they measured the emotions experienced before, during and after  Number 1: node predecessor to node 3 in the path the learning task, using 15 minutes for this. The whole from the source node. experience was organized in a single session.  The subscript 4: step or stage in the solution III. APP DESIGN construction process. RA-AVD is a tool created for learning Dijkstra’s  Strikethrough label (crossing line): it means that the algorithm. The use of the tool is based on the following idea. path denoted by that label is discarded. Node 3 has The student uses the teacher's notes where the Java code that two labels since, at that time, the algorithm had found implements Dijkstra’s algorithm is shown (for the tool to two alternative paths (namely, [3,2]2 and [8,1]4). The recognize it, it has to be a specific source code). When the path with a longer distance from the source to node 3 student has doubts about how the code solves the problem, was discarded. In this case, the label [8,1]4, with he/she uses the tool, focusing on the source code with his/her distance 8, was discarded since the other path was mobile. At that point, the tool shows the source code that the shorter, with length 3. student is viewing on paper, augmenting it with comments and explanations in order to understand how it works. If the student focuses the source code where the graph is declared, RA-AVD tool draws the graph on the mobile’s screen (see "Fixed-Origin Length". Subsequently, the algorithm selects the smallest of these paths (4, 2 and 8), in this case the smallest path is 2, and its corresponding node (node 2), so it is marked as a fixed node for the next stage (step 2). That would be the end of the stage, hence a new stage would start until the minimal path to all nodes is reached. At the top of the trace table, the "Optimal route" and "Length" fields show, at each stage, the optimal path and its length respectively for the selected, fixed node. Initially, all nodes are painted in one color and as they are processed in the construction stages their color changes to show which part of the graph is being processed. IV. RESULTS This section presents the statistical analysis of the data obtained during the experience, in which 18 students aged 19 to 22 participated. Eleven students (61.1%) were men and 7 (38.9%) were women. The analysis was performed with the IBM SPSS tool and the Pandas Matplotlib library in Phyton.x. A. Acquired knowledge Table II shows the level of knowledge acquired by the students after the experience. Students had neither previous contact nor knowledge about greedy algorithms at the beginning of the experience. However, at the end of experience they obtained an average score close to 7 out of 10 (specifically 6.95). Table II displays the mean scores per question (maximum 2 points per question) and the total mean. It should be noted that half of the students in the group (50.5%) obtained a score higher than 8, the maximum score being 10 and the lowest 3. Furthermore, we can see that in Fig. 2. Running the algorithm step-by-step in RA-AVD three of the five questions students scored more than 1.7 out of 2 points. Therefore, that the level of knowledge is quite The trace table has the following columns (see Figure 2): satisfactory.  Step: number of the stage of construction of the TABLE II. KNOWLEDGE EVALUATION QUESTIONNAIRE AND SCORES solution. OBTAINED BY QUESTIONS  Fixed node: the candidate node selected by the Number Statement of the task or question selection function at each stage (of all the candidate Select the distance of two adjacent nodes when they 1 1.72 are equal nodes, the one with the shortest path length from the Indicate the sequence of edges that Dijkstra would source node is selected). 2 1.78 calculate from the origin to node X 3 Find the shortest path from vertex X to vertex Y 0.56  Adjacent nodes: the nodes adjacent to the selected Identify which is the predecessor node of node X in node. 4 1.89 the graph. Starting from the origin node, what would be the  Fixed-origin length: for each adjacent node, it 5 predecessor node Not selected for node X in step N. 1.00 indicates its distance from the source node. Total 6.95  Pending lengths: the lengths of the paths from the B. Positive activation emotions source node that have not been selected so far. It is Figure 3 shows the mean of the positive emotions made up of values from the column "Fixed Origin (enjoyment, hope and pride) valued by the students at three Length" that have not been selected. different times: before, during and after the experience of use  Minimum length: the shortest length of the lengths of of the tool. Not all emotions were measured at all times. For the adjacent nodes of the fixed node and the pending example, measuring proud about accomplishment in the task lengths. does not make sense before having done it, or there is no point in measuring the students’ level of hope about what they will Let's see an example to better understand the meaning of learn once they had learnt it. Therefore, in Figure 3 not all the these columns (note that the source node is 0). Step 0 indicates positive variables appear at all times. the initial state. In step 1, the algorithm marks the origin node 0 as a fixed node (in the first step the origin node itself is We can see that the average enjoyment remains roughly chosen as the fixed node), and determines the adjacent nodes, the same throughout the experience (between 3.72 and 3.70). which are nodes 1, 2 and 4, writing them down in the We also found out that hope decreased in students while they "Adjacent Nodes" column. Next, the algorithm determines the did the task from 4.13 to 3.96 (i.e. 4.12%), and that pride felt distance from these nodes to the origin, whose lengths are 4, once students finished the task decreased from 3.86 down to 2 and 8 respectively, and write them down in the column 3.67 (4.92%). Fig. 4. Negative activation emotions Fig. 3. Positive activation emotions It was also observed that students felt worried and The fact that enjoyment remained constant can be anguished about having to deal with too many study materials interpreted as a favorable symptom, since the learning task and about having little time, since the items of the AEQ entailed concepts that were new and difficult to understand for questionnaire related to this aspect of anxiety received the students, and despite this fact they did not stop enjoying highest ratings (items 86, 96, 111 and 132 in Table III). This during learning. The authors wonder whether the use of AR may have partly increased anxiety during the experience, may have been the reason of keeping constant the levels of reaching a high value at the end of the experience compared enjoyment. In relation to this feeling of enjoyment, it should to the beginning (Figure 4). Note that at the end, anxiety was be noted that item 110 of the AEQ questionnaire (“I study the negative emotion that was experienced with the greatest more than necessary because I enjoy it a lot”) obtained the intensity. Therefore, this finding constitutes an important lowest score with an average of 2.89, which indicates that the aspect that should be taken into account in the design and student is not interested in studying more than strictly construction of educational tools in order to reduce this feeling necessary. Note that we numbered the items same way as the for the student. AEQ questionnaire. Rather, it seems that they were interested Besides, we may observe in Figure 4 that the emotional in acquiring new knowledge that seems significant to them, level of embarrassment decreased throughout the experience, judging by the evaluation of question 139 (“I enjoy acquiring which probably means that the student started feeling more new knowledge”), which obtained the highest score (4.33 out confident as he/she moved forward in the learning activity. of 5). Regarding the decrease in pride, it should be noted that the TABLE III. ANXIETY ITEMS AND THEIR RATINGS highest score of the questions of the AEQ questionnaire to Number Description Time measure this emotion is item 135 (“When I excel in my work item AEQ I feel proud”), which had an average of 4.00 out of 5. Thus, I get so nervous that I don't even want to 82 1,83 BEFORE students value the work they do and feel proud of it. It is start studying. 85 When I have to study I start to feel dizzy. 1,78 possible that they did not value the task they had to do When I look at the books that I have yet to (analyzing Dijkstra’s algorithm) as an interesting task and this 86 2,44 read, I feel distressed. could have caused that pride decreased at the end of the I worry about being able to deal with all 96 2,44 experience. my work. While studying, I want to distract myself C. Negative activation emotions 102 2,94 to reduce my anxiety. DURING In relation to these emotions we could observe that the When time is running out my heart begins 111 2,83 to race. average anger decreased during learning and increased when 118 I get tense and nervous while studying. 1,78 the task was finished by 7.65% (see Figure 4). On the other This subject scares me because I can't hand, the level of anxiety increased notably from the 125 1,83 quite understand it. beginning of the experience, being 15.83% higher during the Worrying about not completing the course 132 2,56 experience and 28.5% at the end of it (Figure 4). makes me sweat. I am concerned that I did not understand AFTER 141 3,56 the subject correctly. When I can't keep up with my studies, it 147 3,17 scares me. D. Negative deactivation emotions The results show that the hopelessness of students decreased while the experience with the AR tool was carried out. However, it increased again to the levels of the beginning when the task was finished (Figure 5). The authors cannot explain well the reason for this effect, although they believe that it could be related to the fact that the use of the AR tool The decrease in these negative emotions during the gave them hope of learning while they were using it. experience could be related to the use of the tool, while the increase detected once the task was finished could be related to the fact that they had to take a knowledge assessment test at the end of the task. This could make them anxious, angry, or hopeless. F. Students’ opinion It was observed that students preferred to work collaboratively during the task. They spontaneously came together in pairs and in some cases even in groups of three students. Some of the students pointed out the lack of any collaborative interaction support in the tool. On the other hand, one of the problems that was observed is that some students did not have a smartphone with minimal characteristics and this caused that the camera focus was not optimal, causing the student to be bringing the mobile closer to the paper several times until the tool detected the source code. The students commented that they were surprised to learn a new subject matter through the tool and even more so Fig. 5. Negative deactivated emotions because some did not know the augmented reality technology Regarding boredom, it was detected that it decreased and it caused them great interest and curiosity, since they had while using the tool (Figure 5). This is an important aspect not commonly used it in the classroom. In general, they since boredom leads to reduced intrinsic motivation and indicated that they were satisfied with the tool and that it had cognitive withdrawal from the task [27]. The authors think been useful in the experience. that the use of augmented reality could have awaken the level V. CONCLUSIONS AND FUTURE WORK of attention and interest of the student and therefore could have reduced boredom, perhaps increasing student’s This article has presented a classroom experience to learn motivation. greedy algorithms using an augmented reality tool called RA- AVD. In this experience, both the emotions and feelings of the E. Comparison of positive and negative emotions students and the level of knowledge acquired have been Figure 6 shows the emotional levels grouped by positive evaluated. The knowledge evaluation was satisfactory. The activation (enjoyment, hope and pride), negative activation students managed to obtain an average of almost 7 (6.95) out (anger, anxiety and embarrassment) and negative of 10, which means that most of the students managed to deactivation emotions (hopelessness and boredom). It should understand the operation and behavior of the algorithm under study (Dijkstra’s algorithm). Note that the students had no be noted that the positive activated emotions experienced by previous knowledge nor contact with greedy algorithms, thus the students as a whole are almost twice as intense as the the learning curve was high. negative ones. The authors wonder if it could be that the use of AR in learning algorithms has something to do with this In relation to emotions experienced by using the tool, the fact of experiencing positive emotions more intensely than experience revealed that the students experienced positive negative ones. Which raises an interesting line of future emotions (enjoyment, hope and pride) much more intensely research. Regarding negative emotions (both activation and than negative ones (such as anxiety or anger), being almost deactivation), it could be seen that they decreased slightly twice as intense. In addition, boredom and embarrassment while students were using the tool in the task (Figure 6, decreased notably during the use of the tool, keeping the “During” section of the diagram). However, at the end all feeling of enjoyment roughly constant during its use. these emotions increased. However, the student's level of anxiety increased from the beginning while using the tool, being almost the double at the end of the experience. This last aspect is especially relevant and it should be taken into account in the design and construction of future educational tools. As future projects, we plan to analyze in greater depth the results obtained by carrying out a correlation analysis between emotions and learning. Furthermore, we intend to replicate the experience with a control group and compare traditional learning with the use of augmented reality. ACKNOWLEDGEMENTS This work has been supported by research grants iProg (ref. TIN2015-66731-C2-1-R) and e-Madrid-CM (ref. P2018/TCS-4307) with FSE and FEDER funds. The support of the GIIAR group of the Salesian Polytechnic University is appreciated. Fig. 6. General average of emotions according to time REFERENCES Workload analysis for task-based, augmented reality-supported robot control. In 2016 25th IEEE International Symposium on Robot and [1] Velázquez-Iturbide, J. Á. The design and coding of greedy algorithms Human Interactive Communication (RO-MAN), pp. 179-184. August revisited. In Proceedings of the 16th Annual Joint Conference on 2016. Innovation and Technology in Computer Science Education, ITiCSE’11 (pp. 8-12). June 2011. [21] Deng, X., Jin, Q., Wang, D. and Sun, F. ARCat: A Tangible Programming Tool for DFS Algorithm Teaching. In Anonymous [2] McConnell, M. Emociones en educación: Cómo las emociones, Proceedings of the 18th ACM International Conference on Interaction cognición y motivación influyen en el aprendizaje y logro de los Design and Children. ACM, pp. 533-537, 2019. estudiantes. Revista Mexicana de Bachillerato a Distancia, 11(21), 2019. [22] Dass, N., Kim, J., Ford, S., Agarwal, S., & Chau, D. H. Augmenting coding: Augmented reality for learning programming. In Proceedings [3] Tapia, A. A. F., Anchatuña, A. L. A., Cueva, M. C., Poma, R. M. M., of the Sixth International Symposium of Chinese CHI (pp. 156-159). Jiménez, S. F. R., & Corrales, E. N. P. Las neurociencias. Una visión April, 2018. de su aplicación en la educación. Open Journal Systems en Revista: Revista Órbita Pedagógica ISSN 2409-0131, (61-74), 2018. [23] Toledo, J. A. J., Collazos, C. A., Cantero, M. O., & Redondo, M. Á. Collaborative strategy with augmented reality for the development of [4] Benavidez, V., & Flores, R. La importancia de las emociones para la algorithmic thinking. In Iberoamerican Workshop on Human- neurodidáctica. Wimblu, 14(1), 25-53, 2019. Computer Interaction (pp. 70-82). Springer, Cham. April, 2018. [5] Zhang, J. Teaching Strategy of Programming Course Guided by [24] McAllister, William. Data structures and algorithms using Java. Jones Neuroeducation. In 2019 14th International Conference on Computer & Bartlett Publishers, 2008. Science & Education (ICCSE) (pp. 406-409). IEEE, August 2019. [25] Pekrun, R., Goetz, T., Frenzel, A. C., Barchfeld, P., & Perry, R. P. [6] Lacave, C., Velázquez-Iturbide, J. Á., Paredes-Velasco, M., & Molina, Measuring emotions in students’ learning and performance: The A. I. Analyzing the influence of a visualization system on students' Achievement Emotions Questionnaire (AEQ). Contemporary emotions: An empirical case study. Computers & Education, 149, educational psychology, 36(1), 36-48, 2011. 103817, 2020. [26] Pekrun, R., Goetz, T., Titz, W., & Perry, R. P. Academic emotions in [7] Serrano-Cámara, L. M., Paredes-Velasco, M., Alcover, C. M., & students' self-regulated learning and achievement: A program of Velazquez-Iturbide, J. Á. An evaluation of students’ motivation in qualitative and quantitative research. Educational Psychologist, 37(2), computer-supported collaborative learning of programming concepts. 91-105, 2002. Computers in Human Behavior, 31, 499-508, 2014. [27] Anaya-Durand, A., & Anaya-Huertas, C. ¿Motivar para aprobar o para [8] Sánchez, M. D. R. G., Añorve, J. R., & Alarcón, G. G. Las Tic en la aprender? Estrategias de motivación del aprendizaje para los educación superior, innovaciones y retos/The ICT in higher education, estudiantes. Tecnología, Ciencia, Educación, 25(1), 5-14, 2010. innovations and challenges. RICSH Revista Iberoamericana de las Ciencias Sociales y Humanísticas, 6(12), 299-316, 2017. [9] Rios, M. D. G., & Paredes, M. Augmented reality as a methodology to development of learning in programming. In International Conference on Technology Trends (pp. 327-340). Springer, Cham, 2018, August. [10] Fidan, M., & Tuncel, M. Integrating augmented reality into problem based learning: The effects on learning achievement and attitude in physics education. Computers & Education, 142, 103635, 2019. [11] Kim, J., Agarwal, S., Marotta, K., Li, S., Leo, J., & Chau, D. H. Mixed Reality for Learning Programming. In Proceedings of the 18th ACM International Conference on Interaction Design and Children (pp. 574- 579). June, 2019 [12] Magnenat, S., Ben-Ari, M., Klinger, S., & Sumner, R. W. Enhancing robot programming with visual feedback and augmented reality. In Proceedings of the 2015 ACM conference on innovation and technology in computer science education (pp. 153-158). June, 2015. [13] Guenaga, M., Menchaca, I., de Guinea, A. O., Dziabenko, O., García- Zubía, J., & Salazar, M. Serious Games, Remote Laboratories and Augmented Reality to Develop and Assess Programming Skills. In International Simulation and Gaming Association Conference (pp. 29- 36). Springer, Cham. June, 2013. [14] Mesia, Natali Salazar, Cecilia Sanz, and Gladys Gorga. "Augmented reality for programming teaching. Student satisfaction analysis." 2016 international conference on collaboration technologies and systems (cts). IEEE, (pp. 165-171) 2016. [15] Teng, C., Chen, J. and Chen, Z. Impact of augmented reality on programming language learning: Efficiency and perception. Journal of Educational Computing Research, 56(2), 254-271, 2018. [16] Melcer, E. Moving to Learn: Exploring the Impact of Physical Embodiment in Educational Programming Games. In Anonymous Proceedings of the 2017 CHI Conference Extended Abstracts on Human Factors in Computing Systems. ACM, pp. 301-306, 2017. [17] Klopfenstein, L., Andriy Fedosyeyev, and Alessandro Bogliolo. "Bringing an unplugged coding card game to augmented reality." Proceedings of the 11th International Technology, Education and Development Conference (IATED). (pp. 9800-9805), 2017. [18] Jin, Q., Wang, D., Deng, X., Zheng, N. and Chiu, S. AR-Maze: a tangible programming tool for children based on AR technology. In Anonymous Proceedings of the 17th ACM Conference on Interaction Design and Children. ACM, pp. 611-616, 2018. [19] Tan, K. S. T. and Lee, Y. An Augmented Reality Learning System for Programming Concepts. In Anonymous International Conference on Information Science and Applications. Springer, pp. 179-187, 2017. [20] Stadler, S., Kain, K., Giuliani, M., Mirnig, N., Stollnberger, G., & Tscheligi, M. Augmented reality for industrial robot programmers: