Enhancing STEM Learning through Interactive Digital Twins: Architecture of a Low-Cost Web Solution for Robotics

Abstract

Educational robotics is a powerful approach for developing STEM skills; however, its widespread adoption is often constrained by the high cost of commercial robotic kits and by the complexity of integrating physical hardware with virtual learning environments. In the context of Industry 4.0, the Digital Twin (DT) concept emerges as a strategic bridge between physical assets and virtual models, yet its application in basic and higher education remains limited by technical and financial barriers. Purpose: To address these challenges, this article presents a web-based Digital Twin architecture that utilizes the low-cost mobile educational platform URA (One Robot per Student) as a case study. The study aims to provide an accessible and scalable solution for real-time monitoring, simulation, and remote interaction, democratizing access to advanced robotics and IoT concepts. Methods: The proposed system integrates an ESP32-based physical robot with a web interface implemented using React.js and Three.js. Communication between the physical and virtual layers is achieved via the MQTT protocol to ensure low-latency data exchange. The platform includes a three-dimensional visualization module for state monitoring and an augmented reality component to support immersive interaction. The research methodology involved a technical evaluation focused on communication latency and synchronization fidelity between the robot’s sensor data (ultrasonic, vibration, and magnetometer) and its virtual counterpart. Furthermore, the study analyzes the pedagogical alignment of the platform with constructivist and constructionist learning principles, emphasizing the importance of active experimentation. Results: Technical evaluations indicate that the system maintains responsiveness and synchronization accuracy within acceptable limits for educational applications, with average latency levels suitable for real-time user perception. The orientation error observed between the physical and virtual entities remained minimal, ensuring a high-fidelity representation. Additionally, the estimated cost of the essential electronic components is approximately US$24.72, which is substantially lower than that of commercial educational robotics platforms, such as Lego Mindstorms. This financial accessibility allows for the replication of the project in resource-constrained institutions. Conclusion: These results demonstrate that the proposed Digital Twin constitutes a viable and cost-effective solution for educational robotics. By enabling students to safely experiment in virtual environments before validating their solutions on real hardware, the platform effectively reduces financial and technical barriers in STEM education. The integration of web technologies and open-source hardware reinforces the scalability of the approach, contributing to the democratization of advanced simulation technologies in the teaching-learning process.