LEARNING COMPLEX SYSTEMS IN STEM EDUCATION WITHIN THE STEAMSMART PROJECT: SIMULATIONS, NEURAL NETWORKS, IoT SENSORS AND ENVIRONMENTAL DATA FOR IMPROVING QUALITY OF LIFE IN HUMAN ENVIRONMENTS

nderstanding complex systems is increasingly important in STEM education for addressing contemporary scientific,
environmental, and societal challenges. Within the STEAMSMART educational framework, currently implemented with
gifted secondary school students, participants engage in hands-on exploration of complex systems through computational
modelling, simulation, environmental sensing and data acquisition.
Students develop and experiment with a variety of computational artifacts inspired by physics and complex systems
science, including Daisyworld simulations, artificial life models, Monte Carlo methods, fractal generation, and dynamic
models of chaotic systems. These activities allow learners to explore key concepts such as feedback mechanisms, emergent behaviour, nonlinear dynamics, and probabilistic processes.
To connect simulations with real-world phenomena, environmental data are collected using IoT sensor platforms based on
ESP32, Arduino microcontrollers, micro:bit and Arduino Q devices. These systems enable the acquisition of parameters
such as temperature, light intensity, humidity, and other environmental variables. The resulting data streams can be
used for analysis and as inputs for machine learning experiments, neural network models, and generative AI-assisted
exploration. The Arduino Q platform, with its hybrid MPU/MCU architecture and on-board AI inference capabilities,
further enables edge deployment of trained models directly on the sensing node.
Programming and experimentation activities are carried out using collaborative coding practices, including emerging
approaches such as “vibe coding”, in co-evolution with locally deployed AI systems that support exploration, debugging,
and iterative model development, also running on edge.
By combining computational simulations, IoT-based environmental sensing, and AI-assisted data analysis, this approach
promotes active and experiential learning while providing students with an integrated understanding of complex systems
across physics, mathematics, computation, and environmental science. The STEAMSMART framework aims to provide
a replicable and scalable educational model for introducing interdisciplinary STEM and AI learning experiences centred
on complex systems, environmental data analysis, and real-world problem solving.