STEM and Computing Education Research
About this research line
We develop evidence-based tools and teaching methods for computational thinking, AI, and robotics reaching tens of thousands of students and teachers across Flanders and beyond.
Introduction
At the AI and Robotics Lab (AIRO), we improve computer science and STEM education through a combination of educational research and classroom practice.
Our goal is simple: help learners build strong digital skills, and help teachers teach these skills with confidence. To do this, we design learning activities, evaluate them in real educational settings, and use the results to continuously improve our materials and tools.
What we research
Computational thinking
We study how computational thinking can be integrated in meaningful ways across the Flemish educational context. Our work includes defining computational thinking for schools in Flanders, developing practical classroom cases across different subjects, and designing validated assessment tools, including work published in Computers & Education. We also contribute to educational resources, including a book on computational thinking.
Impact of digital systems
Digital technology is not only technical; it also has social consequences. We investigate how both perspectives can be taught together. In this line of research, we examine how technology and societal value can be linked in education, how technical and societal learning goals can reinforce one another, and how this integrated approach influences student motivation.
Automated assessment
Teachers need reliable support when evaluating programming and computational thinking. We therefore investigate how automated systems can support teacher assessment, how LLMs can be used for feedback and evaluation, and how to ensure quality and reliability while keeping meaningful human oversight in the loop.
Ways to teach programming
To understand how students learn programming, we collect both quantitative and qualitative data in multiple educational contexts. Our programming environment logs learner interactions and code evolution, allowing us to study learning strategies in depth.
In our Create or Fix experiment, we compared creating programs from scratch with debugging incorrect programs. We found clear behavioral differences: debugging tasks often led to less code editing, but more program execution runs.
Outreach
Social robot competition
Our social robot competition invites learners to design, build, and program robots that interact with people in meaningful ways. It combines creativity, engineering, and computational thinking in authentic, motivating challenges.

Noxi bear

Daggou the robot dog.

Toothbrushing robot BOB

Task robot Tikki
WeGoSTEM
WeGoSTEM introduces primary school children to the key components of robotics: mechanics, electronics, and software. The focus is on hands-on discovery and early engagement with STEM.

The WeGoSTEM drawing robot
Dwengo
Dwengo is an important valorization partner of AIRO. Together, we develop and maintain teaching content and tools, and organize teacher training.
Tools we build
To ensure high-quality research data and strong classroom relevance, we develop our educational tools in-house. These include both hardware and software for physical computing, together with research-based instructional designs validated in real classrooms. All materials are shared with the education community free of charge under a Creative Commons license. Examples include WeGoSTEM, AI at School, and the social robot projects.
Dwenguino Simulator
Programming physical systems offers clear pedagogical benefits. To support this, we created a graphical programming environment for the Dwenguino microcontroller platform. Because Dwenguino is modular and classroom-oriented, learners can program many different physical systems. Our environment also includes a simulator, so students can first focus on programming concepts before dealing with hardware constraints.