ABSTRACT. This paper presents the web-based platform Machine Learning with Bricks and an accompanying two-day course designed to teach machine learning concepts to students aged 12 to 17 through programming-free robotics activities. Machine Learning with Bricks is an open source platform and combines interactive visualizations with LEGO robotics to teach three core algorithms: KNN, linear regression, and Q-learning. Students learn by collecting data, training models, and interacting with robots via a web-based interface. Pre- and post-surveys with 14 students demonstrate significant improvements in conceptual understanding of machine learning algorithms, positive shifts in AI perception, high platform usability, and increased motivation for continued learning. This work demonstrates that tangible, visualization-based approaches can make machine learning concepts accessible and engaging for young learners while maintaining technical depth. The platform is freely available at https://learning-and-dynamics.github.io/ml-with-bricks/, with video tutorials guiding students through the experiments at https://youtube.com/playlist?list=PLx1grFu4zAcwfKKJZ1Ux4LwRqaePCOA2J.

ABSTRACT. This paper presents an educational robotics activity in which lower-secondary students learned about Machine Learning by training, testing and deplyoing a neural network model in a LEGO Spike robot. The didactical goal of this activity is two-fold: propose a learning activity in which students design and build an AI tool (instead of using ready-to-use commercial AI services) and motivate the use of a text-based programming language with a real-world task. Using LEGO SPIKE Prime, students collected RGB sensor data, trained a neural net- work on the robot hub, evaluated model performance, and deployed the trained model to control robot behavior. All pipeline stages were executed on-device, and all students successfully completed the workflow. Pre-post questionnaires examined perceived competence in robot programming, understanding machine learning, and collecting training data. Results showed statistically significant increases across all outcomes, with comparable gains for girls and boys. Students initially unmotivated to learn text-based programming showed significant motivation increases when Python was introduced as a functional tool for embedded systems. This study illustrates the educational potential of on-device machine learning in robotics-based activities for introducing AI concepts at the lower-secondary level.

ABSTRACT. The persistent shortage of STEM professionals necessitates

early and inclusive student engagement. However, many existing out-

reach programs rely on self-selection, often missing students with no prior

interest. To address this gap, we present a school-affiliated internship for

grades 7–10 designed to minimize entry barriers. The curriculum cen-

ters on a low-cost mobile robot and introduces two distinct paradigms:

AI-based line following using a single-layer perceptron and rule-based

maze navigation. This hands-on approach requires no prior knowledge

and emphasizes iterative problem-solving. Observations suggest that the

format can engage a heterogeneous student population. The resulting

concept is resource-efficient and reproducible. It is presented here as a

descriptive curriculum design and observational report for broad-impact

STEM education.

ABSTRACT. In this case study we investigate the integration of institutional AI tools —specifically Gemini Gems and NotebookLM —into a freshmens course at Slovak University of Technology in Bratislava to mitigate the passive use of generative models. We implemented a framework consisting of a Socratic tutor with strict constraints (such as a no-code policy) for exercises and a Retrieval-Augmented Generation (RAG) tool for students preparation to an exam. We think that institutionally managed AI tools can effectively function as external regulators to enforce active learning, with students showing a distinct preference for the grounded reliability of NotebookLM over general chatbots for high-stakes exam revision.

ABSTRACT. This paper focuses on how quality interactions and feedback help maximise the learning potential of educational robots. It forms the final review in a series of published articles examining how robot learning activities benefit from Assessment for Learning (AfL). In robot-mediated activities, quality interactions and feedback occur in real time, as teachers interpret students’ responses and interact with learners while the lesson unfolds. Drawing on constructionist ideas, particularly Papert’s account of learning as the reorganisation of existing mental models, interactions and feedback operate as process-es embedded in activity rather than a discrete instructional act. Educational robots make these processes visible by externalising students’ thinking and supporting dialogue, questioning, and scaffolding. The paper emphasises the central role of teacher judgement, knowledge of learners, and sensitivity to emotional and contextual factors in effective feedback practice. It concludes with brief comments on the potential role of artificial intelligence in supporting, rather than replacing, teacher judgment.

ABSTRACT. The effective integration of Artificial Intelligence (AI) into Vocational Educa-tion and Training (VET) remains challenging due to student heterogeneity and conceptual abstraction. This paper evaluates an AI-oriented robotics curricu-lum developed within the AIM@VET project, consisting of twelve vertically structured Teaching Units. The sequence progresses from basic mobile robot-ics to complex manipulation, visual servoing, and optimization, utilizing phys-ical platforms (Robobo, Dobot Magician) and simulation tools (RoboDK). Adopting a Design-Based Research approach, the study employs a multi-instrument triangulation strategy to assess technical performance, conceptual gains, transversal skills, and teacher feedback across two independent cohorts. Results demonstrate a coherent progression: students maintained stable per-formance in foundational units and successfully consolidated advanced compe-tences, such as vision-based control and algorithmic optimization, despite in-creased cognitive demands. Transversal skills, including debugging and team-work, remained consistently strong. While conceptual assessments confirmed meaningful learning gains, they also identified specific challenges in mathe-matical abstraction. Overall, the findings provide empirical evidence that a progressively scaffold-ed curriculum can support advanced AI learning in VET contexts under realis-tic constraints. This study offers a transferable model and practical insights for educators seeking to integrate AI-driven robotics into vocational programmes.

ABSTRACT. This comprehensive systematic review examines 100+ highly cited peer-reviewed studies (1968-2025) investigating educational robotics interventions in pre-kindergarten settings and their impacts on early literacy and numeracy development. The chronological analysis reveals three developmental phases: the Foundational Period (1968-2006) established theoretical frameworks through Papert's constructionism and Wing's computational thinking; the Empirical Validation Phase (2007-2015) demonstrated pre-kindergarten robotics feasibility with significant computational thinking gains (Cohen's δ=0.27-1.84); and the Integration Era (2016-2025) emphasized curriculum refinement and literacy-numeracy connections. Educational robotics improves sequencing, spatial reasoning, computational thinking, problem-solving, and executive function in children ages 3-6. However, critical gaps persist longitudinal literacy/numeracy tracking, direct evaluation of reading fluency or mathematical operations, broad population representation, and cost-effectiveness analysis remain scarce. Screen-free platforms like KIBO, Bee-Bot, and Roamer demonstrate promise when integrated with structured curricula. Results support strategic, evidence-based integration into pre-kindergarten programs, particularly the U.S. Pre-Kindergarten system, contingent upon rigorous outcome evaluation and professional development investment. Future research must prioritize direct measurement of literacy and numeracy outcomes, longitudinal tracking through kindergarten entry, diverse population representation, and implementation of science approaches to address identified gaps.

ABSTRACT. This short message reports on an ongoing project developing a curriculum-aligned robotics lesson for secondary school students (grade eight and above). The 90‑minute module is implemented as a regular classroom lesson and aims to provide learners with an application‑oriented introduction to robotics through a combination of simulation, exploration, and playful learning. The learning objectives are aligned with the Saxon state curriculum, and the evaluation focuses on both acceptance and the achievement of the intended learning outcomes.

The module addresses fundamental concepts of robotics by explaining the structure and functionality of a robotic arm and enabling students to control and program its movements within a Cartesian coordinate system. Implemented as a laptop-based learning environment, it supports self‑paced, guided exploration, either individually or in small groups. Gamification elements and a purposefully designed interface aim to enhance engagement and accessibility. The simulated robotic arm functions as a digital twin of a collaborative robot, reproducing its motion and functionality within realistic physical constraints.

The evaluation examines students’ motivation, interest, enjoyment, perceived learning gains, fit to individual learning pace, teacher support, and comprehensibility of the content. Initial results from 27 completed questionnaires indicate high acceptance and successful achievement of the learning objectives. Students re-ported strong interest and enjoyment, stated that they were able to learn at their own pace, and confirmed that the learning content was presented in an understandable way.

ABSTRACT. Teaching robotics to computer science students is challenging due to the field's interdisciplinary nature and a common lack of prior experience with hardware and embedded systems. This paper presents a mobile robotics course designed specifically for computer science undergraduates with minimal robotics experience. The course employs a flipped classroom model with interactive web-based materials featuring simulations and visualizations to bridge abstract mathematical concepts with robot behavior. Students build differential-drive robots from off-the-shelf components based on an ESP32-S3 microcontroller, learning prototyping, electronics, and programming skills. The curriculum progresses from high-level programming in Toit to Arduino C/C++, covering topics from concurrency to kinematics and motion planning. All course materials, interactive diagrams, hardware designs, and software are freely available online.

ABSTRACT. With the constant evolution and introduction of new technologies in elementary classrooms in the United States, it is helpful to look back on the origins of the con-struction-based robotics in the classroom to examine how these programs were in-troduced to both teachers and students. When kits such as LEGO TC Logo were in-troduced into education in the 1980’s, there weren’t any previous resources to go by. Papert and the LEGO Group developed their own pedagogy, building on Vygotsky and Piaget. Through a lot of trial and error, teacher training, and expert marketing, the labs grew in popularity. These labs concentrated on academic skills, primarily in math, as well as developing simple programming skills. Eventually, focus shifted, and the use of construction-based robots became more competition-oriented, and many of the basic pedagogical skills that were originally taught were left behind. Today, elementary school labs are utilizing not only kit-based robots, but are introducing 3D printers, laser engravers, and other plug-and-play computer numeric machines (CNC). By looking back, perhaps we can assist teachers in incorporating these new technologies along with the latest generation of technology tools, such as Raspberry Pi, Arduino, and Artificial Intelligence onto the elemen-tary robotics classroom.

ABSTRACT. In this paper, we present ongoing research toward integrating robotics education into existing curricula and study programs. Based on previous research on exploring the possibilities of robotics education in a wider range of educational entities as well as methods of benchmarking, we now explore how to integrate robotics education into general computer science curricula.

ABSTRACT. The integration of robotics and Artificial Intelligence (AI) into education is gaining increasing importance given the ever growing presence of these technologies in both professional and everyday life. To better understand the needs, barriers, and challenges of educators across different European countries and regions regarding robotics and AI classroom integration, this paper presents a systematic study using a mixed-methods approach of combining online surveys and semi-structured interviews with 109 participants from four European countries. The study revealed high interest among educators alongside persistent challenges, including limited experience, lack of suitable hardware and teaching materials, funding and maintenance constraints as well as insufficient integration into curricula. The results provide the empirical basis for general recommendations and the upcoming design of an open-source educational robotics platform, the development of curricula-aligned, ready-to-use teaching materials as well as targeted teacher trainings and support measures in combination with pupil engagement activities in Estonia, Bosnia and Herzegovina, Serbia and Austria.

ABSTRACT. Educational robotics platforms are increasingly used to introduce artificial intelligence concepts to K–12 students, yet most machine learning (ML) workflows still rely on external computers due to hardware limitations. This paper proposes μ-Learn: an open-source, light-weight, fully on-device ML library for educational robots, enabling data collection, training, evaluation, and inference to run entirely on a LEGO® SPIKE™ Prime. The system implements a modular supervised learning workflow from scratch in MicroPython, including dataset handling, multilayer perceptron models, backpropagation, and performance metrics, without any external library. We demonstrate that compact ML models can be trained and deployed autonomously on a severely constrained educational robotics platform. The usage of the library was validated in a middle-school classroom activity where students completed an end-to-end color classification, ML-based workflow. The proposed open-source framework provides a practical foundation for research and experimentation in on-device learning for educational robotics.

ABSTRACT. We report on a newly introduced Human-Robot Interaction course within the Robotics and Intelligent Systems curriculum at Constructor University, Bremen. The course is tightly integrated in a research driven setting of the recently established Robot Interaction Lab. We present the course structure, its embedding within the curriculum, and showcase select studies conducted by the students as part of the course. We include a discussion of experiences from the first iteration of this course.

ABSTRACT. This paper reports the effectiveness of primary school children at predicting likely robot failures that may eventuate during a programming activity session of the Kaspar humanoid robot in a classroom setting. The paper also explores the relationship between the predicted failures, the actual failures that occurred during the programming activity session, and the impact of these predicted and actual failures on the children's enjoyment of the activity. We found that children could not accurately predict those failures of the robot which did organically occur during the session, but neither the predictions nor the failures affected their level of enjoyment.

ABSTRACT. Human–Robot Interaction (HRI) education has largely emphasized technical skills, while psychological and socio-cognitive dimensions of interaction remain underrepresented. Yet contemporary HRI research increasingly views robots as social actors that shape perception, behavior, and decision-making. To address this gap, this paper introduces the Scientific HRI Didactic Framework (SHRI-DiF), a didactical approach that integrates research-based learning with the 5E cycle (Engage–Explore–Explain–Elaborate–Evaluate). The framework positions students as active researchers who conduct the full empirical process, from identifying research questions to evaluating findings. Social robots serve as epistemic tools and experimental agents to investigate social and psychological constructs such as trust, perceived control, and acceptance. The framework is demonstrated in a 12-week undergraduate seminar in business and media psychology using the Furhat robot for dialog strategies in HRI. Results indicate that authentic HRI research settings foster scientific reasoning, methodological competence, and interdisciplinary thinking, offering a transferable model for research-based higher education.