ITU / About ITU / Press / News from ITU / The chair that knows it's a chair

The chair that knows it's a chair

In a study recently published in Nature Communications, researchers at the IT University of Copenhagen show how intelligent building bricks can recognize the objects they form and detect damage.

Written 25 September, 2026 12:12 by Jari Kickbusch

Perhaps you are sitting on a chair while reading this. Now, imagine that your chair is made of electronic building bricks. It looks a bit like a Lego construction, except that each brick of your chair has chips that allow it to communicate with neighbouring bricks. The communication between the bricks creates a kind of collective awareness, which means that
the bricks that form one of the chair's legs are aware that they are part of a chair and that the structure will no longer function properly if they are damaged.

The research article, Smart cellular bricks for decentralized shape classification and damage recovery, offers a glimpse of the future of intelligent objects. In short, it presents a novel system of physical 3D bricks — simple cubic units equipped with local communication, processing, and sensing — that can infer their shape and detect structural damage.

The intelligent chair
The article is the result of a research project, led by researchers from IT University of Copenhagen together with collaborators from Autodesk Research and Sakana AI. It shows how hundreds of simple electronic cubes can work together without any central controller or a map of the structure they form.

Associate professor, Andres Faina, and Assistant professor, Rodrigo Moreno Garcia have co-authored the article, which was recently published in Nature Communications. They have built a small chair to demonstrate how the system works, but the same principles could be used to create many different structures.

"Let's call it smart materials. It's a bit like playing with Lego bricks; you can build whatever you want, but the idea here is that each brick can communicate with the neighbouring bricks. So, if one brick gets damaged or removed, the other bricks will notice," Andres Faina says.

Rodrigo Moreno Garcia explains:

"If some user removes half of this chair, the structure can detect the damage and say; okay, I'm missing a module here. The user can then put the modules back on until the structure itself recognizes that it has been repaired. So, the structure knows where to reconstruct itself if something has happened. It's really, really cool."

Inspired by biological cells
The researchers trained the bricks using a machine learning approach called Neural Cellular Automata, which is inspired by the way biological cells communicate and self-organize. Every brick runs the same neural network and continuously exchanges information with its neighbours. Over time, these local interactions allow the bricks to develop a shared understanding of the overall shape.

"Basically, it's an adaption of how our bodies works. It's like a model of some of the functions we have in our brain and how our brains connect with our muscles," Andres Faina says.

Perspective
In the study, the approach proved highly successful. In simulations, the system correctly classified shapes nearly 99 percent of the time. The team then transferred the software to physical hardware, building structures containing nearly 200 bricks. In every test, the bricks successfully agreed on the correct shape.
Andres Faina and Rodrigo Moreno Garcia emphasize that the current system is still a proof of concept. The bricks require external power and cannot yet physically rearrange themselves. However, the study shows that the underlying learning framework scales well to larger and more complex structures with large amounts of components.
Looking ahead, the technology could contribute to future generations of adaptive materials, modular robotics, autonomous construction systems, and damage-tolerant infrastructure. In the shorter term, it could also enable educational systems and smart building materials that provide local feedback about structural integrity and assembly progress.
"So far, our work remains largely at the laboratory stage, and I don't expect we'll see intelligent IKEA chairs anytime soon. The more immediate potential may lie in safety-critical systems where reliability is essential. We are also exploring applications in modular robotics, where robots can transform into different shapes and adapt to changing environments." ends Andres Faina.

Learn more:
Smart cellular bricks for decentralized shape classification and damage recovery, by Rodrigo Moreno, Andrés Faíña, Shyam Sudhakaran, Kathryn Walker and Sebastian Risi, published in Nature Communications (2026). The research was conducted by the IT University of Copenhagen in collaboration with Autodesk Research and Sakana AI.