Next Article in Journal
Teacher Beliefs and Perspectives of Practice: Impacts of Online Professional Learning
Previous Article in Journal
Educational Mobile Apps for Programming in Python: Review and Analysis
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Education Sciences
Volume 13
Issue 1
10.3390/educsci13010067
education-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Teaching Machine Learning in K–12 Using Robotics

by
Georgios Karalekas
,
Stavros Vologiannidis
* and
John Kalomiros
Department of Computer, Informatics and Telecommunications Engineering, International Hellenic University, 62124 Serres, Greece
*
Author to whom correspondence should be addressed.
Educ. Sci. 2023, 13(1), 67; https://doi.org/10.3390/educsci13010067
Submission received: 11 November 2022 / Revised: 26 December 2022 / Accepted: 30 December 2022 / Published: 10 January 2023
(This article belongs to the Section STEM Education)
Versions Notes

Abstract

:
Artificial intelligence (AI) and machine learning (ML) are pursued in most fields of data analysis, and have already become a part of everyday applications. As AI and ML are an integral part of the Industry 4.0 era, it becomes necessary to introduce a basic understanding of what AI and ML means and how it can be applied in K–12 education. Although educators need to prepare for this revolution, it is generally admitted that there is a shortage of suitable tools and methods toward this goal. In this article, we propose that it is necessary to design courses for machine learning using STEM-based robotic tools. We present selected robotic kits with ML capabilities that can be used to target concepts of machine learning in K–12 classrooms. Finally, we present our own conceptual rules on how constructivist educational robotics with AI capabilities can effectively support teaching scenarios in future K–12 curricula.

1. Introduction

During the past few years, artificial intelligence (AI) has entered the daily lives of non-experts by its availability in everyday devices such as mobile phones. AI’s influence has already reached all parts of society where data are available and data collection is possible [1]. The tremendous increase in computing power in the last twenty years has allowed applications to successfully deliver results that were previously impossible. One of the most successful fields of AI is machine learning (ML), which includes techniques that allow computers to learn from data [2]. ML processes are employed in many widely used devices and services. For example, when tagging photos on social media, ML is used to detect faces. When interacting with speech-based personal assistant services, ML is used for speech recognition. Τhe targeting of advertisements in various social networks is also based on machine learning techniques. Heys [3] argues that the advances in ML techniques and applications will reshape work, education, entertainment, and even human relations. As ML services become increasingly ubiquitous, understanding the underlying ML processes is important for the average user. However, despite the pervasive nature of ML, very few understand the technology behind it [4]. This lack of understanding creates a misplaced fear for artificial intelligence and automation, overshadowing their potential positive impact on society. The position of Heys [3] is in line with the Royal Society Technical Report [5], which emphasizes that an understanding of basic ML processes is necessary for everyone, regardless of age, especially for children. Children will be the future consumers of such services, but they will also be the future scientists, who will further develop and advance ML tools and methods. It has been reported that AI start-ups need more new talent for their endeavors [6,7]. Therefore, it is important for governments and organizations to help citizens, and especially digital natives, to develop a basic understanding of artificial intelligence technologies [8].
The 4th Industrial Revolution is directly affecting the global workforce as it requires workers with new skills [9,10,11]. In this context, P21’s Frameworks for 21st Century Learning, suggests and defines in [12] the following skills and knowledge needed by students to succeed in work and life: creativity and innovation, critical thinking and problem solving, communication, and collaboration. The above educational goals are served by the well-known STEM initiative (Science, Technology, Engineering, Math) [13,14,15] in modern education. STEM has been complemented by STEAM (Science, Technology, Engineering, Arts, Math) [16], where art lessons are also supported. The concept of STEM is an integrated form of curriculum, information, and assessment to enhance science education by seamlessly combining these four disciplines [17]. STEM education has begun to gain ground over traditional ways of learning in recent years. It has been confirmed that constructivist learning is more understandable and equips students with more skills [18]. STEM education can play a crucial role in the sustainable development of a country’s economy and stability by creating critical thinkers and the next generation of innovators [19].
In recent years, educational robotics (ER) has attracted much attention from educators and researchers as a tool to support formal education [20,21], especially STEM and STEAM curricula. The main motivation for introducing ER tools into classrooms is based on Seymour Papert’s construct [22]. In this context, educational robots appear to be a predefined embodiment of learning artifacts [15,22]. It has been argued that ER promotes student interest in STEM disciplines [21,23,24,25,26] and can be used to impart technical competencies such as programming skills [27,28,29]. Educational robotics is an effective tool for improving student learning: students are engaged in a fun and hands-on learning environment while they use constructivist learning approaches, where the focus is on the learning process rather than the end product [30]. Through involvement with robots, children can learn, in a subtle way, concepts that are traditionally obscure. Ιt has been shown that by working with ER, students can acquire important transversal skills such as critical thinking, problem solving, decision making, communication, or teamwork [15,31,32,33]. Furthermore, previous work has recognized the potential of robotics to promote the development of computational thinking skills [34,35]. Other studies have shown that ER can have positive effects on the students’ motivation, self-confidence, and creativity [36,37], thus facilitating a happier way of learning. Robots have become essential in STEM courses as they can combine all areas of STEM methodology [18,23,26,38].
Bearing in mind the success of educational robots in educational scenarios, we see an opportunity to introduce the concept of machine learning through educational robots. As part of our study, first, we present the current state in teaching ML in K–12, then we introduce a collection of commercial robots that can readily be used by teachers in teaching ML concepts, and finally, we conclude with a new proposition for using manipulators as the preferred ER platform for introducing ML.
The rest of the article is organized as follows. In Section 2, the motivation in using robotics in AI and ML lessons is presented. In Section 3, we lay the criteria of our investigation regarding the selection of AI and ML tools and robots for the classroom. In Section 4, the results on the existing tools and robots for teaching ML are presented. In Section 5, we put forward some concepts on teaching AI and ML through AI powered educational robotics, dictated by our own experience. Section 6 concludes the paper.

2. Motivation

The AI4K12 initiative of the Association for the Advancement of Artificial Intelligence (AAAI) and the Computer Science Teachers Association (CSTA) of the United States of America introduced the five AI’s “Big Ideas” [8], with the third one being that computers can learn from data through machine learning. According to this, students should understand that machine learning is a type of statistical inference that finds patterns in data without being pre-programmed to do so. In another study conducted at the “Teaching Machine Learning Workshop” [39], which targeted those who would like to know how teachers from around the globe approach teaching ML, we found that the basic knowledge students should acquire through courses on machine learning fall into four categories:
  • Everyday literacy (knowledge based on personal and communal experience);
  • Applied literacy (skill-based literacy: using a specific skill of know-how, based on acquired expertise);
  • Theoretical literacy (disciplinary knowledge);
  • Reflexive literacy (probing assumed and specialized knowledge systems).
In order to investigate the students’ attitudes toward ML, and in a preliminary effort to identify how ML should be taught in K–12, the platform “Machine Learning for Kids” [40] was used in two school classes with 9th graders. Experimental lessons were conducted, with the goal of teaching students how to train a computer to recognize faces. Although the topic initially interested the students, 11 out of 41 students (a percentage of 26.8%) did not put in the required effort to deliver the final project. Some students explained their loss of interest due to the fact that the same functionality was included in their mobile phone. This is in contrast to previous experiences with the same students who were involved in twelve STEM lessons and none of them gave up. Students, especially younger ones, tend to have difficulties in fully understanding abstract concepts and ideas without applying them to real-life situations, where physical objects are manipulated.
Based on the above reactions, we realized that by using ER for teaching machine learning and applying the concepts in real life situations, we could make these abstract concepts visible and concrete and thus the lessons successful. Although robots have been the companion of students in STEM lessons, there have been very few attempts to teach machine learning through STEM educational approaches in which robots are used [41,42,43,44].
Regarding everyday literacy, by using robots in real situations, the students can better understand both the limitations and possibilities of ML. They can also identify possible applications where ML is not necessary. In such cases, traditional programming methods can have better results and work faster, for example, a line-following application [45].
Toward the goal of applied literacy, students can see the results of their work applied to actual circumstances. Constructing robotic artifacts and making them move successfully is more engaging and pleasing than watching an algorithm executing on a computer screen [22,46]. Additionally, by applying supervised/unsupervised ML methodologies to robots, students receive a hands-on understanding of the different ways machine learning works. By applying unsupervised reinforcement learning when, for example a robot is learning to move forward using its arms [47,48] or supervised ML when a robot learns to drive on a simulated road after receiving training data [49] can be very beneficial for students.
With regard to theoretical literacy, students can try to customize and integrate algorithms into code. They can also learn to integrate useful ML libraries in order to implement robot tasks in real-world physical problems. This may help students to understand some of the hidden complexities of ML.
Finally, regarding probing further knowledge acquisition, students can build upon their systems by expanding and reprograming their intelligent robots.

3. Methodology

With the above in mind, we provide an overview of educational robots that include some out of the box AI capabilities, without depending on other platforms to gain AI skills. Our focus was on robots that can be programmed in a language suitable for students and that also include examples and training material in their documentation. Finally, the chosen robots were small enough to be suitable for use in an average school classroom. We did not restrict robots according to their ML capabilities. ML models range from simple color recognition to more complex ones that can be trained to recognize faces, emotions, sounds, and poses.
Additionally, we provide an overview of existing ML teaching applications in K–12 and the related literature, in order to stress the different readiness levels between teaching ML using traditional (virtual) tools and educational robots.

4. Results

4.1. Existing Tools for Teaching ML

The fundamentals of ML concepts and techniques have been introduced in higher education institutions (HEIs) as part of university curricula. However, young children who grow up and interact with smart devices such as Alexa have no contact with ML as a concept. Therefore, extending computing education to include ML concepts as early as possible and providing teachers with the necessary teaching material represents a new challenge [50,51].
There are several tools available for the introduction of ML, focusing at the early stages of K–12 education. For example, AI for Oceans [52,53], which is designed for grade three and up, introduces ML, training data, and ML bias while aligning with some concepts of the United States’ Computer Science Teachers Association Computer Science Standards [54]. Additionally, there are AI/ML-enhanced online tools and resources that introduce AI and ML, trying to be as fun for the use of children as possible. For example, there are several AI/ML-tools accessible to K–5 including Scratch Lab’s Face Sensing Scratch extensions [55,56] for text to speech, translation, video sensing, and face sensing [57]. Other online tools include Machine Learning for Kids [40,58], and Cognimates [59,60]. These efforts have been addressed toward experienced teachers and consider specific teaching environments with a small number of students. Therefore, they are difficult to adapt to real school setups. One of the platforms that have been created for the dissemination of ML is the Teachable Machine, which uses ML classification models without the need of specialized prior knowledge [61]. Similarly, LearningML [62] was designed to teach ML by allowing students to create their own applications, while Zhorai [63] was designed to help children explore ML and knowledge representation. Additionally, some curricula such as PopBots [64,65] and IRobot [66] have been designed to introduce artificial intelligence into secondary K–12 education, but have only been implemented experimentally.
It has been reported that the lack of comprehensive AI curricula in core K–12 courses has become one of the barriers to introducing artificial intelligence in the classroom [67]. It is worth mentioning that only China has officially started systematic teaching in some schools in the country [68].
Despite the above observation, several steps have been taken for the introduction of ML concepts to young students during the past few years [69,70]. A systematic study mapping the state-of-the-art on teaching machine learning in elementary and high schools showed that there has been a rapid increase in instructional units (Ius) developed in the last years [71]. The study showed that most of the Ius have been proposed as extracurricular units ranging from 1-h taster workshops to semester-long courses, but they are not included in the school curriculum. The study in [72] charted the emerging trajectories in educational technology, theory, and practice, related to teaching machine learning in K–12 education, and reached the conclusion that teaching ML in K–12 posed an even more daunting challenge than struggling to integrate computational thinking into school curricula. In [73], they suggest that despite the challenges, there is a clear need for an early understanding of how ML-based and data-driven systems can solve real-world problems.

4.2. AI-Powered Educational Robotics

In the following, we present robotic tools that abide to the criteria presented in the Section 3. The selected robots have ready-made modules that allow them to be trained to recognize faces (facial recognition), persons (face detection), emotions, sounds, objects, colors, and hand gestures. Arduino robotic KITS or BBC micro:bit were not included since they do not provide ML capabilities out of the box. In order to acquire such capabilities, these kits must be used in conjunction with other platforms or must be programmed in languages difficult for young learners to understand. In the following, we focused on six stand-alone, AI-powered educational robots: Zumi, RoboMaster S1, ClicBot, Cozmo, MINDSTORMS Robot Inventor, and Cogbots. Some of their properties are presented in Table 1. The columns of the table are explained below:
  • Robot: The name of the robot;
  • Type: The shape and the type of the robot;
  • Control unit: The central control unit of the robot;
  • Actuators: The output units of the robot;
  • Sensors: The input units of the robot;
  • Open source: The code is open source or not;
  • Coding languages: The scripting languages for controlling the robot;
  • Training: The AI/ML modules that can be trained from the user;
  • Curriculum: Lesson plans, instructor units, and useful material for educational use of the robots.

4.2.1. Zumi

Zumi [74] is a tiny buildable self-driving car kit. Its focus is to help students learn how autonomous cars use sensors and cameras to navigate around the world and learn about the environment using AI. The central control unit is a Raspberry Pi0 and has six infrared sensors, a gyrometer, an accelerometer, a Pi camera, and a 128 × 64 OLED screen. It can be programmed using Blockly or Python. It is able to make decisions based on color detection, hand gestures, and face recognition using ML. It allows the student to program reactions when a sign is identified, or when a person is recognized. Additionally, Zumi can be trained to identify objects and make decisions accordingly. The company that makes it offers free lesson plans for ages 6 to 12+ on their website [74].

4.2.2. RoboMaster S1

The RoboMaster S1 educational robot [75], is an educational robot including AI technology. It can combine training in programming skills with solving problems in mathematics and physics. RoboMaster S1 supports Scratch and Python programming languages. The Controller can simultaneously support functions such as low-latency high-definition image transmission, AI computing, and program development. The S1 features four Mecanum wheels, each with 12 cylinders that allow for omnidirectional movement and precision control. AI technology allows the S1 to recognize gestures, sounds, and even other S1 robots and perform tasks such as automated driving. It can recognize roads and traffic lights and can perform line recognition, clap recognition, person recognition, and gesture recognition. With vision marker recognition, it can recognize up to 44 vision markers featuring numbers, letters, and special characters. A series of projects, video tutorials, and programming guides ranging from beginners to experts have been developed. Τhe robot is suitable for ages 10 to 16 up, while some information about the training material can be found on the company’s website [75].

4.2.3. ClicBot

ClicBot [76], is a robotic kit designed for kids with built-in functions for education and entertainment with AI capabilities. It has a modular design that can be assembled and disassembled. This allows children to create all sorts of imaginative robots such as building bricks. The Brain is the master control and power supply unit of the ClicBot. The eye of the ClicBot is a 2.1-inch rotational, circular touch screen that contains a camera with AI facial recognition software, gesture sensors, and camera. There are three touch sensors located on the upper, left, and right side of the Brain, respectively, and two connectors are located on the upper and bottom rear side. The Brain is integrated with four functional modules:
(a)
The Joint—used for integrated motion;
(b)
The Skeleton primarily used for building limbs;
(c)
The Wheel used for vehicle setups;
(d)
The Smart Foot that includes a highly sensitive pressure sensor and a micro-processor to control and measure terminal pressure, etc.
Clicbot is programmable with a drag-and-drop coding interface based on Blockly by Google and is also compatible with Python. Τhe robot is suitable for ages 6 to 16+. Information about the training/educational material can be found on the company’s website [76].

4.2.4. Cozmo

By writing programs for various missions for the Cosmo robot [77], students can discover the concept of artificial intelligence. Cozmo is equipped with proximity sensors, a gyroscope, a downward-facing cliff detector, and a camera that allows it to learn human faces and objects and sense human emotions. It also has a screen and in combination with a speaker, it expresses emotions. Cozmo comes with three LED cubes, which are used in games that test the reaction time and color matching. It can learn how you play using ML and adjust its own skill level and reactions appropriately. It can be programmed with Code Lab, a block coding environment built on Scratch Blocks for younger students and with Python for older. Code Lab is a coding app that runs on a tablet. Calypso [78] is another coding app for Cozmo, which also works on tablets. Calypso is a simple tile-based user interface to teach robot logic and behavior. The Cosmo provides SDKs for experienced developers to explore the possibilities of AI. It also provides an AI course that encourages students to learn how AI works, while developing coding skills. The robot is suitable for ages 6 to 16+. Information about the training/educational material can be found on the company’s website [77].

4.2.5. MINDSTORMS Robot Inventor

Lego in August 2022 introduced machine learning in a programming environment called LEGO® MINDSTORMS® Robot Inventor App [79] that can be used with the MINDSTORMS Robot Inventor Kit [80]. This particular programming platform is based on the logic of Scratch but also supports programming in Python language. The kit includes the brain with an integrated gyroscope, four medium-size motors, one color sensor, one distance sensor, and can also accept a force sensor. The student can make different robots depending on the application they want to create. Machine learning is an extension to the application that uses the camera or microphone on the device. The robot can be programmed to recognize different objects or voice commands. The application can be installed on computers, Android tablets, and mobile phones. Lego provides some examples of building robots along with their programming code. One can also find plenty of instructions on the Internet for building various robots along with their code. The robot is suitable for ages 8 to 16+. Information about the training/educational material can be found on the company’s website [80].

4.2.6. Cogbots

Cogbots [81], exists in two versions, CogBot and CogMini, which are open-source robotics kits developed by CogLabs in collaboration with Google and UNESCO. The robot’s skeleton is 3D printed and each student can create their own robot using simple everyday materials and recycled old smartphones. For the controller, CogBot uses an Arduino ESP32 board connected with motors and a smartphone. It is worth noting that CogBot uses the smartphone’s sensors. Its motors are 360-degree continuous rotation micro servo motors, which are low-cost, and help keep the robot’s price affordable. The smartphone communicates with a PC computer for programing via the local Wi-Fi network and with the ESP32 board via Bluetooth. Cogbots can be programmed with Scratch. The smartphone does not need a mobile data subscription as long as it is Android 5.0 or later. Teachable Machines, from Google Creative Lab, is an add-on tool that allows CogBot to provide machine learning tools [82] for controlling the AI powered robot, so that students learn AI concepts. The ThinkBot Scratch extension allows students to use an image classification model created with Teachable Machine in their codes to control CogBot. Using the Scratch extension, AI technology is accessible to students with little or no coding experience. Τhe robot is suitable for ages 8 to 16+. Information about how to set up and control the robot can be found on the website [81].

5. Some Conceptual Guidelines for Teaching ML with STEM-Based Robotics

Employing the constructivist STEM approach using educational robots is very promising in teaching scenarios for machine learning. Additionally, the educational robots presented above are innovative products, but they also give rise to some considerations concerning their use in the classroom. First, they are mobile, which means they need space for their movement, especially the largest among them, the RoboMaster S1. Therefore, it is difficult to use them in an ordinary classroom, without much space for their movement. Most of their software is not open source, which limits the development of new lesson plans adapted to the students. All robots need a wireless connection to a computer and a Wi-Fi network. Finally, the machine learning models are not customizable and as a result, the learner cannot experiment. All of the above may hinder how the courses are conducted. The only robot that can avoid the above restrictions is the Lego MINDSTORMS Robot Inventor. This is a kit that can be configured as each student or teacher wants and can also be connected with a cable to a computer.
To overcome some of the above problems, we propose the use of educational robotic arms as teaching devices for teaching ML. The main reasons are given below:
  • The robotic arms are static and do not need a lot of space.
  • Unlike a robotic car, the robotic arm does not need a wireless connection and it can be next to the student, connected via an USB cable.
  • A static robot arm is much more manageable by the teacher in the classroom.
  • Programming a robotic arm for ML tasks is more meaningful than programming a mobile robot.
Several examples can be given to support point D. An interesting ML problem is training the robotic arm to do simple movements such as in pick and place tasks. In another task, the robot can learn to recognize images, words, or even poses and react accordingly. Finally, by using a robotic arm, the concept of a collaborative robot, a human assistant robot, is more understandable to children.
We have inferred some conceptual guidelines that can help the configuration of teaching scenarios for ML:
Use low-cost materials and recycle existing parts. A good example is Cogbots.
  • Limit the workspace of the robot, in order to make it more fit for the classroom.
  • Robots do not necessarily need to have a powerful controller as long as they can connect to a computer.
  • Regarding connection types, the use of a USB cable should be preferred instead of a wireless network, which are prone to problems.
  • Create short teaching scenarios to accommodate the lesson in the given timeslot. Use object or sound recognition in combination with specific robot movement. Try to design a lesson plan where the robot is trained to become the student’s assistant. This connection will create a sense of familiarity, which will motivate the student.
  • Leave room for experimentation by using open parameters such as the number of neural network hidden layers in TensorFlow Playground [83] or the learning parameters in a Q-Learning algorithm. Configuring the parameters allows for a better understanding of the underlying ML concepts as will using a static robotic arm rather than a mobile robot.

6. Conclusions and Future Work

Contemporary education requires the introduction of artificial intelligence, especially machine learning to be introduced in schools. Through STEM and robotics, these concepts can be made more tangible and understandable for 21st century students. To be able to easily use robots in the classroom, it is more convenient for the robots to not be mobile but static such as robotic arms. Robotics kits must be of relatively low cost for a school to afford multiple robots. Programming environments should enable students to be able to experiment with machine learning but without having to engage in difficult mathematical concepts. Finally, robotic kits should enable teachers to take advantage of unplanned teachable moments and deviate from standard lesson plans and innovate, listening to the needs of their students.
For the upcoming school year, the authors have designed and will implement lesson plans that will introduce the concept of artificial intelligence into secondary education classes using robotic arms. In the initial lessons, the manipulators will be used as the students’ robotic assistants, who will be trained with supervised learning. In the following lessons, the robotic arms will use reinforcement learning to learn how to perform a certain movement. The specific courses will include pre- and post-assessment tests. The results will be compared to last year’s results of corresponding courses on introducing the concepts of machine learning, which were conducted by the authors and did not use robots.

Author Contributions

Conceptualization, S.V., G.K. and J.K.; Methodology, S.V. and G.K.; Investigation, G.K. and J.K.; Resources, J.K.; Writing—original draft preparation, G.K.; Writing—review and editing, S.V. and J.K.; Supervision, S.V.; Project administration, J.K. All authors have read and agreed to the published version of the manuscript.

Funding

EPAnEk–NSRF 2014–2020, Region of Central Macedonia: ΚΜΡ6-0078707.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. De la Higuera, C. A Report about Education, Training Teachers and Learning Artificial Intelligence: Overview of Key Issues. Working Paper Prepared for the Knowledge Societies Division, Communication and Information Sector of the UNESCO. 2019. Available online: http://cdlh7.free.fr/UNESCO/Teaching_AI-cdlh.pdf (accessed on 11 January 2022).
  2. Alpaydin, E. Introduction to Machine Learning, 4th ed.; MIT Press: Cambridge, MA, USA, 2020. [Google Scholar]
  3. Heys, J. Machine Learning as a Tool to Identify Critical Assignments. Chem. Eng. Educ. 2018, 52, 4. [Google Scholar]
  4. Evangelista, I.; Blesio, G.; Benatti, E. Why Are We Not Teaching Machine Learning at High School? A Proposal. In Proceedings of the 2018 World Engineering Education Forum—Global Engineering Deans Council (WEEF-GEDC), Albuquerque, NM, USA, 12–16 November 2018; pp. 1–6. [Google Scholar] [CrossRef]
  5. Royal Society. Machine Learning: The Power and Promise of Computer That Learn by Example; Technical Report; The Royal Society: London, UK, 2017; ISBN 978-1-78252-259-1. [Google Scholar]
  6. Muro, M.; Whiton, J.; Maxim, R. What Jobs Are Affected by AI? Better-Paid, Better-Educated Workers Face the Most Exposure; Brookings Institution: Washington, DC, USA, 2019. [Google Scholar]
  7. Bessen, J. AI and Jobs: The Role of Demand; Working Paper No. 24235; National Bureau of Economic Research: Cambridge, MA, USA, 2018. [Google Scholar] [CrossRef]
  8. Touretzky, D.S.; Gardner-McCune, C.; Martin, F.; Seehorn, D. K-12 Guidelines for Artificial Intelligence: What Students Should Know. In Proceedings of the ISTE Conference, Philadelphia, PA, USA, 23–26 June 2019. [Google Scholar]
  9. Marr, B. Why Everyone Must Get Ready for the 4th Industrial Revolution. Forbes, 4–6. 2016. Available online: https://www.forbes.com/sites/bernardmarr/2016/04/05/why-everyone-must-get-ready-for-4th-industrial-revolution/print/ (accessed on 18 July 2022).
  10. Marr, B. What Is Industry 4.0? Here’s a Super Easy Explanation for Anyone. Forbes, 4–7. 2018. Available online: https://www.forbes.com/sites/bernardmarr/2018/09/02/what-is-industry-4-0-heres-a-super-easy-explanation-for-anyone/#5b3ba6539788 (accessed on 19 July 2022).
  11. Chatzopoulos, A.; Papoutsidakis, M.; Kalogiannakis, M.; Psycharis, S. Innovative Robot for Educational Robotics and STEM. In Intelligent Tutoring Systems; Kumar, V., Troussas, C., Eds.; Springer International Publishing: Berlin/Heidelberg, Germany, 2020; pp. 95–104. [Google Scholar] [CrossRef]
  12. Battelle for Kids. Framework for 21st Century Learning. Partnership for 21st Century Learning. 2019. Available online: https://www.battelleforkids.org/networks/p21/frameworks-resources (accessed on 15 March 2022).
  13. Alimisis, D.; Moro, M. Special issue on educational robotics. Robot. Auton. Syst. 2016, 77, 74–75. [Google Scholar] [CrossRef]
  14. Barker, B.S.; Nugent, G.; Grandgenett, N.; Adamchuk, V.I. Robots in K-12 Education: A New Technology for Learning; IGI Global: Pennsylvania, PA, USA, 2012; pp. 1–402. [Google Scholar]
  15. Benitti, F.B.V. Exploring the educational potential of robotics in schools: A systematic review. Comput. Educ. 2012, 58, 978–988. [Google Scholar] [CrossRef]
  16. Kim, Y.; Park, N. The Effect of STEAM Education on Elementary School Student’s Creativity Improvement. In Computer Applications for Security, Control and System Engineering; Kim, T., Stoica, A., Fang, W., Vasilakos, T., Villalba, J.G., Arnett, K.P., Khan, M.K., Kang, B.-H., Eds.; Springer: Berlin/Heidelberg, Germany, 2012; pp. 115–121. [Google Scholar] [CrossRef]
  17. Breiner, J.M.; Harkness, S.S.; Johnson, C.C.; Koehler, C.M. What Is STEM? A Discussion About Conceptions of STEM in Education and Partnerships. Sch. Sci. Math. 2012, 112, 3–11. [Google Scholar] [CrossRef]
  18. Takacs, A.; Eigner, G.; Kovacs, L.; Rudas, I.J.; Haidegger, T. Teacher’s Kit: Development, Usability, and Communities of Modular Robotic Kits for Classroom Education. IEEE Robot. Autom. Mag. 2016, 23, 30–39. [Google Scholar] [CrossRef]
  19. Xie, Y.; Fang, M.; Shauman, K. STEM Education. Annu. Rev. Sociol. 2015, 41, 331–357. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  20. Tselegkaridis, S.; Sapounidis, T. Exploring the Features of Educational Robotics and STEM Research in Primary Education: A Systematic Literature Review. Educ. Sci. 2022, 12, 305. [Google Scholar] [CrossRef]
  21. Alimisis, D. Educational robotics: Open questions and new challenges. Themes Sci. Technol. Educ. 2013, 6, 1. [Google Scholar]
  22. Papert, S. Mindstorms: Children, Computers and Powerful Ideas; Basic Books: New York, NY, USA, 1980; Volume 1. [Google Scholar]
  23. Atmatzidou, S.; Demetriadis, S. Advancing students’ computational thinking skills through educational robotics: A study on age and gender relevant differences. Robot. Auton. Syst. 2016, 75, 661–670. [Google Scholar] [CrossRef]
  24. Vollstedt, A.M.; Robinson, M.; Wang, E. Using robotics to enhance science, technology, engineering, and mathematics curricula. In Proceedings of the American Society for Engineering Education Pacific Southwest Annual Conference, University of Nevada, Reno, NV, USA, 12–13 April 2007; pp. 1–9. [Google Scholar]
  25. Mataric, M.J.; Koenig, N.; Feil-Seifer, D. Materials for enabling hands-on robotics and STEM education. In Proceedings of the AAAI Spring Symposium: Semantic Scientific Knowledge Integration, Stanford University, Stanford, CA, USA, 26–28 March 2007; pp. 99–102. [Google Scholar]
  26. Levy, R.B.-B.; Ben-Ari, M.M. Robotics Activities–Is the Investment Worthwhile? In Informatics in Schools. Curricula, Competences, and Competitions; Brodnik, A., Vahrenhold, J., Eds.; Springer International Publishing: Berlin/Heidelberg, Germany, 2015; pp. 22–31. [Google Scholar] [CrossRef]
  27. Alimisis, D.; Kynigos, C. Constructionism and robotics in education. In Teacher Education on Robotic-Enhanced Constructivist Pedagogical Methods; Alimisis, D., Ed.; ASPETE: Athens, Greece, 2009; pp. 11–26. [Google Scholar]
  28. Atmatzidou, S.; Markelis, I.; Demetriadis, S. The use of LEGO Mindstorms in elementary and secondary education: Game as a way of triggering learning. In Proceedings of the International Conference of Simulation, Modeling and Programming for Autonomous Robots (SIMPAR), Venice, Italy, 3–7 November 2008; pp. 22–30. [Google Scholar]
  29. Nourbakhsh, I.R.; Crowley, K.; Bhave, A.; Hamner, E.; Hsiu, T.; Perez-Bergquist, A.; Richards, S.; Wilkinson, K. The Robotic Autonomy Mobile Robotics Course: Robot Design, Curriculum Design and Educational Assessment. Auton. Robot. 2005, 18, 103–127. [Google Scholar] [CrossRef] [Green Version]
  30. Eguchi, A. Bringing robotics in classrooms. In Robotics in STEM Education; Khine, M.S., Ed.; Springer: Cham, Switzerland, 2017; pp. 3–31. [Google Scholar] [CrossRef]
  31. Blanchard, S.; Freiman, V.; Lirrete-Pitre, N. Strategies used by elementary schoolchildren solving robotics-based complex tasks: Innovative potential of technology. Procedia—Soc. Behav. Sci. 2010, 2, 2851–2857. [Google Scholar] [CrossRef] [Green Version]
  32. Atmatzidou, S.; Demetriadis, S.N. Evaluating the Role of Collaboration Scripts as Group Guiding Tools in Activities of Educational Robotics: Conclusions from Three Case Studies. In Proceedings of the 2012 IEEE 12th International Conference on Advanced Learning Technologies, Rome, Italy, 4–6 July 2012; pp. 298–302. [Google Scholar] [CrossRef]
  33. Petre, M.; Price, B. Using Robotics to Motivate ‘Back Door’ Learning. Educ. Inf. Technol. 2004, 9, 147–158. [Google Scholar] [CrossRef] [Green Version]
  34. Repenning, A.; Webb, D.; Ioannidou, A. Scalable game design and the development of a checklist for getting computational thinking into public schools. In Proceedings of the 41st ACM Technical Symposium on Computer Science Education, Milwaukee, WI, USA, 10–13 March 2010; pp. 265–269. [Google Scholar] [CrossRef]
  35. Lee, I.; Martin, F.; Denner, J.; Coulter, B.; Allan, W.; Erickson, J.; Malyn-Smith, J.; Werner, L. Computational thinking for youth in practice. ACM Inroads 2011, 2, 32–37. [Google Scholar] [CrossRef]
  36. Miller, D.P.; Nourbakhsh, I.R.; Siegwart, R. Robots for education. In Springer Handbook Robotics; Springer: Heidelberg, Germany, 2008; pp. 1283–1301. [Google Scholar]
  37. Khanlari, A. Effects of robotics on 21st century skills. Eur. Sci. J. 2013, 9, 26–36. [Google Scholar]
  38. Naya, M.; Varela, G.; Llamas, L.; Bautista, M.; Becerra, J.C.; Bellas, F.; Prieto, A.; Deibe, A.; Duro, R.J. A versatile robotic platform for educational interaction. In Proceedings of the 2017 9th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS), Bucharest, Romania, 21–23 September 2017; Volume 1, pp. 138–144. [Google Scholar] [CrossRef]
  39. Steinbach, P.; Seibold, H.; Guhr, O. Teaching Machine Learning in 2020. In Proceedings of the First Teaching Machine Learning and Artificial Intelligence Workshop, Virtual, 8–13 September 2021; pp. 1–6. Available online: https://proceedings.mlr.press/v141/steinbach21a.html (accessed on 23 April 2022).
  40. Machine Learning for Kids. Available online: https://machinelearningforkids.co.uk/ (accessed on 11 February 2022).
  41. Eguchi, A. Machine Learning Application with AI-Robotics Tool—What We Learned from Pilot Studies. In Robotics in Natural Settings; Cascalho, J.M., Tokhi, M.O., Silva, M.F., Mendes, A., Goher, K., Funk, M., Eds.; Springer International Publishing: Berlin/Heidelberg, Germany, 2023; pp. 383–394. [Google Scholar] [CrossRef]
  42. Dietz, G.; King Chen, J.; Beason, J.; Tarrow, M.; Hilliard, A.; Shapiro, R.B. ARtonomous: Introducing Middle School Students to Reinforcement Learning Through Virtual Robotics. Interact. Des. Child. 2022, 430–441. [Google Scholar] [CrossRef]
  43. Zhang, Z.; Willner-Giwerc, S.; Sinapov, J.; Cross, J.; Rogers, C. An Interactive Robot Platform for Introducing Reinforcement Learning to K-12 Students. In Robotics in Education; Merdan, M., Lepuschitz, W., Koppensteiner, G., Balogh, R., Obdržálek, D., Eds.; Springer International Publishing: Berlin/Heidelberg, Germany, 2022; pp. 288–301. [Google Scholar] [CrossRef]
  44. Eguchi, A. Understanding Machine Learning Through AI-powered Educational Robotics—Pilot Study with Undergraduate Students. In Robotics in Education; Lepuschitz, W., Merdan, M., Koppensteiner, G., Balogh, R., Obdržálek, D., Eds.; Springer International Publishing: Berlin/Heidelberg, Germany, 2022; pp. 52–62. [Google Scholar] [CrossRef]
  45. TY—THES AU—Peters, Lina PY—2019/12/05 SP—T1—An Educational Programming Environment for Teaching the Principles of Machine Learning Using LEGO MINDSTORMS DO—10.13140/RG.2.2.25925.60647 ER. Available online: https://medium.com/swlh/machine-learning-on-lego-mindstorms-ev3-2384004702f9 (accessed on 12 January 2022).
  46. Melchior, A.; Cohen, F.; Cutter, T.; Leavitt, T. More than Robots: An Evaluation of the FIRST Robotics Competition Participant and Institutional Impacts; Heller School for Social Policy and Management, Brandeis University: Manchester, NH, USA, 2005. [Google Scholar]
  47. Tim, T. Reinforcement Learning with Lego Mindstorms—Crawling Robot. Available online: https://youtu.be/bVbT9zkPIvs (accessed on 20 June 2022).
  48. Rousseau, V. Lego Mindstorms: One-Legged Robot Learning How to Crawl with q-Learning. Available online: https://youtu.be/F-kNXAFeffs (accessed on 22 June 2022).
  49. Narahara, T.; Kobayashi, Y. Personalizing homemade bots with plug & play AI for STEAM education. In SIGGRAPH Asia 2018 Technical Briefs; Association for Computing Machinery: New York, NY, USA, 2018; pp. 1–4. [Google Scholar] [CrossRef]
  50. Sulmont, E.; Patitsas, E.; Cooperstock, J.R. What Is Hard about Teaching Machine Learning to Non-Majors? Insights from Classifying Instructors’ Learning Goals. ACM Trans. Comput. Educ. 2019, 19, 33:1–33:16. [Google Scholar] [CrossRef]
  51. Ho, J.W.; Scadding, M.; Kong, S.C.; Andone, D.; Biswas, G.; Hoppe, H.U.; Hsu, T.C. Classroom activities for teaching artificial intelligence to primary school students. In Proceedings of the International Conference on Computational Thinking Education, Hong Kong, China, 13–15 June 2019; pp. 157–159. [Google Scholar]
  52. Code.org, AI for Oceans. Available online: https://code.org/oceans (accessed on 3 May 2022).
  53. Yang, W. Artificial Intelligence education for young children: Why, what, and how in curriculum design and implementation. Comput. Educ. Artif. Intell. 2022, 3, 100061. [Google Scholar] [CrossRef]
  54. Computer Science Teachers Association. CSTA Standards for Computer Science Teachers. 2020. Available online: https://csteachers.org/teacherstandards (accessed on 3 May 2022).
  55. Scratch Lab’s Face Sensing. Available online: https://lab.scratch.mit.edu/face/ (accessed on 5 May 2022).
  56. Aki Tamashiro, M.; Van Mechelen, M.; Schaper, M.-M.; Sejer Iversen, O. Introducing Teenagers to Machine Learning through Design Fiction: An Exploratory Case Study. Interact. Des. Child. 2021, 471–475. [Google Scholar] [CrossRef]
  57. Scratch Extensions. Available online: https://scratch.mit.edu/projects/editor (accessed on 10 May 2022).
  58. Yu, W.J.; Jang, J.H.; Ahn, J.M.; Park, D.R.; Yoo, I.H.; Bae, Y.K.; Kim, W.Y. Design of Artificial Intelligence Education Program based on Design-based Research. Int. J. Adv. Smart Converg. 2019, 8, 113–120. [Google Scholar] [CrossRef]
  59. Cognimates Team. Cognimates. 2020. Available online: http://cognimates.me/ (accessed on 12 May 2022).
  60. Druga, S. Growing Up with AI: Cognimates: From Coding to Teaching Machines. Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA, USA, 2018. [Google Scholar]
  61. Carney, M.; Webster, B.; Alvarado, I.; Phillips, K.; Howell, N.; Griffith, J.; Jongejan, J.; Pitaru, A.; Chen, A. Teachable Machine: ApproachableWeb-Based Tool for Exploring Machine Learning Classification. In Proceedings of the CHI 2020, Honolulu, HI, USA, 25–30 April 2020. [Google Scholar]
  62. Rodríguez-García, J.D.; Moreno-León, J.; Román-González, M.; Robles, G. Introducing Artificial Intelligence Fundamentals with LearningML: Artificial Intelligence made easy. In Proceedings of the Eighth International Conference on Technological Ecosystems for Enhancing Multiculturality, Barcelona, Spain, 26–29 October 2021; pp. 18–20. [Google Scholar] [CrossRef]
  63. Lin, P.; Brummelen, J.V.; Lukin, G.; Williams, R.; Breazeal, C. Zhorai: Designing a Con-versational Agent for Children to Explore Machine Learning Concepts. In Proceedings of the AAAI Conference on Artificial Intelligence, New York, NY, USA, 7–12 February 2020; Volume 34. [Google Scholar] [CrossRef]
  64. Williams, R.; Park, H.W.; Oh, L.; Breazeal, C. PopBots: Designing an Artificial Intelligence Curriculum for Early Childhood Education. In Proceedings of the AAAI Conference on Artificial Intelligence, Honolulu, HI, USA, 27 January–1 February 2019; Volume 33, p. 1. [Google Scholar] [CrossRef] [Green Version]
  65. Williams, R.; Park, H.W.; Breazeal, C. A is for Artificial Intelligence: The Impact of Artificial Intelligence Activities on Young Children’s Perceptions of Robots. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, Glasgow, UK, 4–9 May 2019; pp. 1–11. [Google Scholar] [CrossRef]
  66. Burgsteiner, H.; Kandlhofer, M.; Steinbauer, G. IRobot: Teaching the Basics of Artificial Intelligence in High Schools. In Proceedings of the AAAI Conference on Artificial Intelligence, Phoenix, AZ, USA, 12–17 February 2016; Volume 30, p. 1. [Google Scholar] [CrossRef]
  67. Van Brummelen, J.; Lin, P. Engaging Teachers to Co-Design Integrated AI Curriculum for K-12 Classrooms. arXiv 2020, arXiv:2009.11100. [Google Scholar] [CrossRef]
  68. Jing, M. China Looks to School Kids to Win the Global AI Race. South China Morning Post. 2018. Available online: https://www.scmp.com/tech/chinatech/article/2144396/china-looks-school-kids-win-global-ai-race (accessed on 5 February 2022).
  69. Hitron, T.; Wald, I.; Erel, H.; Zuckerman, O. Introducing children to machine learning concepts through hands-on experience. In Proceedings of the 17th ACM Conference on Interaction Design and Children, Trondheim, Norway, 19–22 June 2018; pp. 563–568. [Google Scholar] [CrossRef]
  70. Vartiainen, H.; Tedre, M.; Valtonen, T. Learning machine learning with very young children: Who is teaching whom? Int. J. Child-Comput. Interact. 2020, 25, 100182. [Google Scholar] [CrossRef]
  71. Marques, L.S.; Gresse Von Wangenheim, C.; Hauck, J.C.R. Teaching Machine Learning in School: A Systematic Mapping of the State of the Art. Inform. Educ. 2020, 19, 283–321. [Google Scholar] [CrossRef]
  72. Tedre, M.; Toivonen, T.; Kahila, J.; Vartiainen, H.; Valtonen, T.; Jormanainen, I.; Pears, A. Teaching Machine Learning in K-12 Computing Education: Potential and Pitfalls. arXiv 2021, arXiv:2106.11034. [Google Scholar] [CrossRef]
  73. Hintz, A.; Dencik, L.; Wahl-Jorgensen, K. Digital Citizenship in a Datafied Society; John Wiley & Sons: Hoboken, NJ, USA, 2018. [Google Scholar]
  74. Zumi. Available online: https://www.robolink.com/products/zumi (accessed on 17 June 2022).
  75. Robomaster S1. Available online: https://www.dji.com/gr/robomaster-s1 (accessed on 10 June 2022).
  76. Clicbot. Available online: https://keyirobot.com/en-eu (accessed on 13 June 2022).
  77. Cozmo. Available online: https://www.digitaldreamlabs.com/pages/cozmo (accessed on 20 June 2022).
  78. Calypso. Available online: https://calypso.software (accessed on 20 June 2022).
  79. Inventor Mindstorms APP. Available online: https://apps.microsoft.com/store/detail/lego%C2%AE-mindstorms%C2%AE-robot-inventor/9MTQ0N7W1D6X?hl=el-gr&gl=gr (accessed on 24 June 2022).
  80. Mindstorms Inventor Kit. Available online: https://www.lego.com/en-gr/product/robot-inventor-51515 (accessed on 24 June 2022).
  81. CogBots. Available online: https://sites.google.com/ucsd.edu/airoboticswithcogbots/home (accessed on 26 June 2022).
  82. Google Teachable Machine. Available online: https://teachablemachine.withgoogle.com/train/image (accessed on 12 April 2022).
  83. TensorFlow Playground. Available online: https://playground.tensorflow.org/ (accessed on 22 February 2022).
Table
Table 1. Table of robot features.
Table 1. Table of robot features.
RobotTypeControl UnitActuatorsSensorsOpen SourceCoding LanguagesTrainingCurriculum
ZumiSelf-driving car kitRaspberry Pi 02 DC motors,
OLED screen.		Gyrometer, accelerometer, camera, 6 IR sensors.NoPython,
Blockly		Identify objects,
learn colors, hand gestures, and faces.		Ready lesson plans.
RoboMaster S1Omni directional self-driving car with blasterNo Info6 brushless motors
Blaster, Gimbal.		Camera,
Microphone,
IR range detectors,
Hit detectors.		NoScratch, Python,
RoboMaster app		Face detection path following, tracking, visual marker, gesture recognition.Video courses.
ClicBotModular robot (4–Kits)No infoDc motors
speaker,
suction cup, Screan, Griper.		Camera,
distance/touch sensors,
gesture sensor, microphone.		No ClicBot App.
Blockly, Python		Face detection,
assign movement directions.		Community
robot models and programs.
CosmoSelf-driving car with small manipulatorNXP Kinetis K02 100MHz ARM Cortex M4Display screen, speaker, 4 DC motors, LEDS.Proximity sensors, gyroscope, a downward-facing cliff detector, camera.NoCode Lab (Scratch), Python, CalypsoFaces, human feelings, objects. Video courses, coding tutorials.
MINDSTORMS Robot InventorModular Robot KitSmart Hub4 servo
motors		Color sensor, distance sensor, gyroscope.NoRobot Inventor App (Scratch),
Python		Image recognition, sound recognition.Robot build
examples and code.
Cogbotsself-driving car kitArduino ESP32,
Smartphone 		2/4 micro servo
motors, Smartphone speaker, screen		Sensors of the smartphone.YesScratch, ThinkBot Scratch extensionDetection faces, objects, sounds, poses.Instructions how to use and set up the robot.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Karalekas, G.; Vologiannidis, S.; Kalomiros, J. Teaching Machine Learning in K–12 Using Robotics. Educ. Sci. 2023, 13, 67. https://doi.org/10.3390/educsci13010067

AMA Style

Karalekas G, Vologiannidis S, Kalomiros J. Teaching Machine Learning in K–12 Using Robotics. Education Sciences. 2023; 13(1):67. https://doi.org/10.3390/educsci13010067

Chicago/Turabian Style

Karalekas, Georgios, Stavros Vologiannidis, and John Kalomiros. 2023. "Teaching Machine Learning in K–12 Using Robotics" Education Sciences 13, no. 1: 67. https://doi.org/10.3390/educsci13010067

APA Style

Karalekas, G., Vologiannidis, S., & Kalomiros, J. (2023). Teaching Machine Learning in K–12 Using Robotics. Education Sciences, 13(1), 67. https://doi.org/10.3390/educsci13010067

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop