Introduction:
The growing interest in introducing concepts of Artificial Intelligence (AI) and Robotics into basic education has fostered the development of didactic tools that are accessible, practical, and pedagogically relevant. This paper presents FRANKIE, a low-cost, open-source educational robot designed to support introductory teaching of AI and robotics in school contexts. The proposal addresses the need for educational resources that combine ease of use, technical feasibility, and the potential to stimulate logical reasoning, creativity, and the understanding of AI algorithm functioning (neural networks) among elementary and secondary school students.

Frankie Robot:
FRANKIE was developed with a modular architecture based on widely available components, such as Raspberry Pi microcontrollers, ultrasonic sensors, DC motors, a camera, and an optional 3D-printed physical structure. This design enables wide reproducibility and customization by educators and students alike. Its AI component is based on the weightless neural network WiSARD (Wilkes, Stonham and Aleksander Recognition Device), allowing the development of activities that introduce fundamental AI concepts, such as reactive behavior programming, sensor-based control, decision-making algorithms, and basic notions of reinforcement learning.

Educational material:
To assess the educational potential of the tool, practical workshops were prepared and tested with students in basic education. Participants were able to assemble, program, and interact with the robot in hands-on activities. The results revealed a high level of engagement and enthusiasm from the students, as well as a satisfactory understanding of autonomous behaviors, neural network functioning, and programming logic.
Furthermore, teachers reported the tool’s ease of integration into interdisciplinary pedagogical proposals encompassing content from mathematics, physics, and computer science (programming and AI concepts). The open-source nature of the project—featuring public access to assembly files, source code, and tutorials—reinforces its inclusive and collaborative character, encouraging educational communities to adapt and expand the use of FRANKIE according to their specific needs. Compared to widely adopted commercial solutions such as LEGO EV3, Thymio, and Bee-Bot, FRANKIE offers a more economically viable alternative without compromising on educational features.

Final remarks:
Future work includes expanding the repository of available didactic materials and conducting longitudinal studies to evaluate the impact of continued use of the platform on the development of computational and cognitive AI-related competencies across different age groups.
In summary, FRANKIE represents a significant contribution to the field of educational robotics and AI teaching, combining accessibility, flexibility, and pedagogical purpose, with strong potential to democratize access to these emerging technologies within the school environment.

Keywords: Artificial Intelligence in education, Neural Network, Educational Robotics, Technologies in education.