ABSTRACT. Robotics courses often struggle to bridge the gap between mathematical foundations and robust real-world robot behavior. Open robotics competitions provide authentic, motivating problem settings, but their artifacts and lessons learned are rarely reused in teaching. This paper proposes a bidirectional approach that links competitions and university robotics education: course exercises produce modular, ROS-based components that students can reuse in later competition participation, while competition outcomes (e.g., robots and game elements, simulation worlds, and scenarios) are converted into sustainable teaching assets. Using Eurobot as a case study, we outline how competition-derived tasks can be decomposed into scaffolded hands-on exercises for an introductory robotics curriculum covering coordinate transformations, kinematics, localization, and navigation.

ABSTRACT. e-Yantra Robotics Competition (eYRC), a pioneering initiative by IIT Bom-bay, Educational Robotics Competitions (ERCs) provide authentic, project-based contexts that foster problem-solving, teamwork, and computational thinking. While prior research has documented the challenges faced by nov-ice teams in such environments, less is known about how proficient teams approach complex robotics tasks. This study investigates the problem-solving strategies of expert teams participating in the e-Yantra Robotics Competition (eYRC) 2022–23, a national educational robotics initiative by IIT Bombay. Using semi-structured interviews with seven finalist teams across distinct ro-botics domains, data were analyzed through a combined deductive–inductive approach, guided by Jonassen’s three-stage problem-solving framework. Findings reveal six interrelated strategies underpinning expert performance: (1) problem understanding and framing, (2) systematic task decomposition, (3) activation of prior knowledge, (4) iterative implementation and debug-ging, (5) collaborative coordination, and (6) effective time and resource management. These strategies collectively reflect high levels of cognitive regulation, metacognitive awareness, and collaborative competence. The in-sights have been translated into the design of PRoboT, a scaffolded web-based learning environment aimed at explicitly teaching structured problem-solving to future eYRC participants. This work contributes to understanding expert cognition in robotics education and provides a foundation for devel-oping pedagogical scaffolds that cultivate similar competencies among nov-ice teams.

ABSTRACT. This paper presents an instructional framework based on the 5E Model (Engage, Explore, Explain, Elaborate, Evaluate) that integrates Artificial Intelligence (AI)-based marker recognition into aerial robotics programming. Designed to promote STEAM engagement and engineering thinking, it moves beyond basic piloting to involve students in real-world problem-solving with block-based and Python coding, introducing them to the fundamentals of AI vision. Grounded in project-based learning, the framework demonstrates how accessible aerial platforms can enhance interdisciplinarity. Moreover, It addresses a notable gap in the educational robotics literature, offering a scalable pedagogical guide that provides educators with principles and strategies for adapting learning experiences in autonomy and intelligent systems.

ABSTRACT. Integrating autonomous robots naturally into human environments remains a significant challenge. Magic performances provide an effective testbed for human-robot interaction, requiring real-time perception, precise manipulation, and natural communication with audiences. We present an autonomous humanoid robot that performs the interactive coin trick “Which hand is the coin in?” without any human assistance. In our system, we propose the Crossed Arm Voxel Network (CAVN), a deep learning architecture comprising three 3D convolutional layers that classify 32×32×32 voxelized LiDAR point clouds to determine human arm positioning. CAVN was trained on 2,000 labeled 3D scans, achieving 94% validation accuracy. We deployed and tested CAVN on our modified THORMANG3 robot (upper body only, 27 DOF with 12 per arm) during the IROS 2019 magic and music competition, where it won first place. The overall system is distributed across three computers integrated through the Robot Operating System. The system achieved over 90% accuracy in real-time classification during live audience interaction, demonstrating that humanoid robots can autonomously perform complex interactive tasks.

ABSTRACT. Unibotics is an open web framework for robot programming and teaching robotics in Higher Education. It is based on ROS, supports both Gazebo simulated robots and real ones, and includes several courses on Mobile Robotics, Service Robotics, Computer Vision in Robotics, etc. This paper describes a brand new Intelligent Tutoring System (ITS) that has been integrated into Unibotics to help students in their learning process and increase their autonomy during laboratory work. It is based on a ChatBot that answers student questions following a Retreival-Augmented Generation (RAG) pipeline, with Qwen2.5 Large Language Model (LLM) and the robotics user documentation available in the platform as domain specific knowledge base. Three case studies have been extensively tested and the ITS has shown good answers and conversations.

ABSTRACT. We present pyCub, an open-source physics-based simulation of the humanoid robot iCub, along with exercises to teach students the basics of humanoid robotics. Compared to existing iCub simulators (iCub SIM, iCub Gazebo), which require C++ code and YARP as middle-ware, pyCub works without YARP and with Python code. The complete robot with all articulations has been simulated, with two cameras in the eyes and the unique sensitive skin of the iCub comprising 4000 receptors on its body surface. The exercises range from basic control of the robot in velocity, joint, and Cartesian space to more complex tasks like gazing, grasping, or reactive control. The whole framework is written and controlled with Python, thus allowing to be used even by people with small or almost no programming practice. The exercises can be scaled to different difficulty levels. We tested the framework in two runs of a course on humanoid robotics. The simulation, exercises, documentation, Docker images, and example videos are publicly available at https://rustlluk.github.io/pyCub.

ABSTRACT. In this paper, we unify, refactor, expand on, and refine previous implementations to map motion capture data to simulated and real robot executions. We introduce an integrated and general reinforcement learning framework that is usable for undergraduate students in robotics courses who extended the implementations with specialized observations and reward functions to instruct complex robots. The framework covers the entire pipeline of motion capture and refinement, re-targeting procedures, mapping for dynamic feasibility, Sim2Sim, and Sim2Real. The implementation of the framework is generic wrt. the chosen motion capture, simulation, and robot systems.

ABSTRACT. As AI becomes increasingly embedded in everyday life, children often anthropomorphize these systems, attributing emotions and moral understanding to what are essentially trained algorithms. Our goal with this study was to inves- tigate how middle school students' perceptions of AI evolve through hands-on robotics workshops using the AlphAI platform. Seven sixth-grade students par- ticipated in three structured 45 minutes sessions focused on neural network com- plexity, sensor integration, and data bias, while building and training LEGO ro- bots. Pre- and post-questionnaires, combined with qualitative interviews, re- vealed nuanced shifts in students' understanding of AI reliability, responsibility attribution, and the role of training data. Students struggled to connect workshop activities with real-world AI systems like ChatGPT, exhibited ownership bias when evaluating their robots' performance, and failed to spontaneously recognize the need for multiple sensors in complex tasks. These findings highlight critical gaps between hands-on manipulation and conceptual transfer, offering insights for designing more effective AI education interventions.

ABSTRACT. To prepare students for upcoming trends and challenges, it is important to teach them about the helpful and important aspects of modern technologies, such as robotics. However, classic study programs often fail to prepare students for working in the industry because of the lack of practical experience, caused by solely theoretical lecturing. The challenge is to teach both practical and theoretical skills interactively to improve the students’ learning. In the scope of the paper, a project- based learning approach is proposed, where students are taught in an agile, semester-spanning project how to work with robots. This project is part of the applied computer science degree study program Digital Technologies. The paper presents the framework as well as an exemplary project featuring the development of a disassembly software ecosystem for hardware robots.

ABSTRACT. Hands-on robotics workshops can foster early interest and confidence in STEM, yet classroom use is often hindered by fragile hardware, unintuitive assembly, and short runtimes from limited battery capacity and inefficient power management. This paper presents a redesigned, open-source version of the Otto DIY robot featuring USB-C charging and programming, improved energy management, and durable mechanics with metal-gear servos. The bill of materials is approximately 54 USD per unit, enabling students to take their builds home while keeping programs affordable at scale. It was developed and deployed over multiple years during the “Schülerinnen-Projekttage” (Girls’ Project Days) of the Technische Universität Darmstadt, a workshop series for female secondary school students designed to engage participants in STEM through hands-on activities. Aggregated pre- and post-surveys from 2022–2025 for the Otto-based workshop (n ≈ 82) show students’ self-perceived technical confidence doubled and interest in studying STEM increased from 37% to 63%. In practice, the redesign reduced downtime from assembly errors and power-related interruptions, decreasing observed frustration and improving workshop flow. These results suggest that child-friendly design choices can translate limited contact time into more meaningful, creative learning experiences and sustain interest in STEM.

ABSTRACT. We present the design, implementation, and verification of a vision-based localization method for indoor flight of autonomous drones. The method identifies square-corner features of a reference mat in images captured by an onboard, gimbal-stabilized, downward-facing camera and enables reliable estimation of the drone pose without external infrastructure. Based on this localization approach, we developed a software system for synchronized flight of a drone swarm following predefined choreographies. The system supports reusable trajectory segments that can be executed relative to an arbitrary initial drone position and orientation, making it well suited for educational and demonstrational use. The complete solution has been deployed as a permanent public installation in the SteelPark science center in Košice, where it is regularly demonstrated to school groups and summer camp participants. All software components are released as open source and are available online.

ABSTRACT. We describe a novel masters-level projects class that teaches robotics along the traditional robotics pipeline (dynamics, state estimation, controls, planning). One key motivational part is that students have to directly apply the algorithms they learn on a highly constrained compute platform, effectively making a robot fly. We teach nonlinear algorithms as deployed in state-of-the-art flight stacks such as PX4. Didactically, we rely on two core concepts: 1) avoidance of provided black-box software infrastructure, and 2) usage of the safe and efficient programming language Rust that is used on the PC (for simulation) and an STM32 microcontroller (for robot deployment). We discuss our methodology and the student feedback over two years with ten students each.

Teaching material: https://imrclab.github.io/teaching/flying-robots

ABSTRACT. Robotics education increasingly demands scalable and interdisciplinary learning models capable of addressing both foundational concepts and real-world applications. In response to this need, the Robotics Goes MOOC project, aimed at broadening access to robotics science and technology, was conceived to present robotics through four complementary paradigms—knowledge, design, interaction, and impact. This paper presents the design and pedagogical rationale of the Robotics & Robots MOOC’s program, developed as the educational core of the project and delivered on the Federica Web Learning platform. The MOOC program is structured as a two-course pathway, Robotics in a Nutshell and Robots in Action, each comprising eleven lectures delivered by multiple internationally renowned experts. Starting from the four paradigms of the Robotics Goes MOOC project, the program implements a curriculum that integrates foundational robotics principles, embodiment and interaction, application domains, as well as societal and ethical considerations. Attention is given to topic sequencing, coordination of multi-instructor contributions, and the balance between conceptual depth and accessibility for learners with heterogeneous backgrounds. We present the preliminary results from an initial survey conducted to gather sociodemographic data from over 200 participants enrolled in the program courses, including information on age, educational qualification, professional category, sector of employment, and gender.

ABSTRACT. Unibotics is an open web framework for robot programming and teaching robotics in Higher Education. It is based on ROS and supports both Gazebo simulated robots, real ones, and recently also real cameras and video files. This paper describes the new available Computer Vision in Robotics course, which includes 8 state-of-the-art practical challenges for undergraduate and graduate level students. They allow hands-on learning of visual perception, deep learning for visual perception, visual localization and vision based decision making, including the visual control of a mobile robot using end-to-end deep learning. This course is also available at the offline fully open-source version of Unibotics, named RoboticsAcademy.

ABSTRACT. Educational robotics is a new field of research that focuses on the development of robotics systems capable of autonomously navigating their environment. These systems are based on a combination of artificial intelligence (AI) and machine learning (ML) algorithms, enabling the creation of autonomous robots capable of multitasking that are also affordable and energy-efficient, which is crucial for their deployment in schools and industry. In this paper, we will compare several state-of-the-art edge cameras and the functionalities they offer. We will also include our own experience with their use.

ABSTRACT. This paper presents Emio, a platform that combines a reconfigurable parallel continuum robot with Emio Labs, which provides a simulation-based digital twin and a set of ready-to-use novel education labs (or exercises), to support the dissemination of soft robot technology. The Emio robot has 4 rotating motors (which provide the robot’s controlled degrees of freedom), a depth camera for sensing, and a system of interchangeable soft legs and a set of end-effectors giving different functions to the robot like inspection and pick and place. The proposed labs cover important concepts for soft robotics such as: modeling of deformation, direct and inverse kinematics, mechanical design of the legs and closed-loop control. The paper presents the platform, a unified simulation approach, and comprehensive educational content enabling students and young researchers to understand fundamental concepts of modeling in soft robotics. The proposed educational contents, occasionally unprecedented, are examined in the context of their potential for advancing knowledge sharing.

ABSTRACT. In this paper, Pico4Drive is described, which is a custom development board for small robot prototyping. In order to highlight its benefits, a differential robot prototype, based on the Pico4Drive development board, will be compared with a recently prototyped Arduino based small robot, which was a very successful and trendy technology in recent years. The described mobile robot prototype was assembled to be applied in educational contexts, having DC motors with built-in encoders, ToF sensors and an analog line sensor, being its mechanics built by applying 3D printing technology.

ABSTRACT. Aligned with National Educational Policy 2020 of India and UN SDG 4, P.K.M. College of Education, a Government Aided Teacher Education Institution affiliated to Kannur University initiated STEM Education to strengthen teacher competencies for the 21st Century. STEM Educational Initiatives provides hands-on training in robotics, coding, electronics, and 3D printing, through industry–academia collabora-tion and institutional STEM infrastructure. Structured across pre-internship, intern-ship, and post-internship phases, the initiative enables student teachers to serve as STEM mentors in schools while enhancing their pedagogical and technological skills. The initiatives have resulted in increased student engagement, recognition at science fairs, and participation in international robotics competitions, demonstrating its effectiveness as a scalable model for future-ready STEM teacher education, par-ticularly in a rural context.

ABSTRACT. This paper examines the pedagogical impact of developing an autonomous mobile robot designed for precision agriculture in community greenhouses. Developed through an interdisciplinary collaboration between mechanical and software engineering students, the project follows a Software-First ap-proach, where mechanical design is dictated by AI-driven perception re-quirements. The system integrates a Raspberry Pi 5 with a Hailo-8 AI accel-erator to perform real-time YOLOv8-based detection of crop diseases at 20-30 FPS. By bridging the gap between a 2.5-credit IoT course and a capstone software project, the initiative provides a robust Project-Based Learning (PBL) framework. We discuss how challenges in Edge AI computing, cloud-based CAD collaboration (Onshape), and stakeholder engagement foster es-sential systems-engineering competencies and CDIO skills. The project demonstrates high educational value by connecting advanced engineering education with environmental sustainability and community-driven needs.