Buildings that Reshape Themselves

The challenge

Our environment is constantly changing. We need our built environment to respond to those changes to remain relevant and effective. We’re working with the University of New South Wales to explore just that. Here are just two ways we’re looking to design buildings that are responsive, adaptive, and incorporate the best of artificial intelligence, robotics, machine learning and computational design.

In this article

  • An inflatable, origami inspired meeting room capable of interacting with the people inside it.
  • Macro-structures, made up of synchronised robot swarms.
  • Soft robotics, biomimicry and the trend influencing modern robotics.
  • What all this has to do with future of the built environment.

Our environment is constantly changing. By the hour, as the sun makes a long sweeping arch across the sky. By the day, as seasons pass and temperatures rise and fall. Other changes are less cyclical and more permanent — sea levels rise, devouring coast lines and earthquakes re-carve our topography. Why then does our built environment, often designed as a response to our natural environment, remain static? How much of the stagnant nature of our buildings is a result of technological limitations? How much — given the maturity of technology like artificial intelligence, robotics, and machine learning — is simply a result of status quo thinking?

This is a question, we’re trying to answer collaboratively with students and academics from the University of New South Wales (UNSW). We’re working with UNSW’s Computational Design (CoDE) program on two projects that explore dynamic and responsive architecture. When we say dynamic, we mean the architecture is capable of changing its form and function over the duration of its lifetime. By responsive, we mean it’s capable of recognising the need to do so.

One project looks at how we can use modern and emerging technologies to create rooms that can learn from human behaviour and change their shape in response to how they are used.

The project was proposed by Associate Professor M. Hank Haeusler, the Discipline Director of the UNSW’s CoDe program. Hank and his colleague, Associate Lecturer Alessandra Fabbri, wanted to explore how six emerging fields—artificial intelligence and machine learning, digital fabrication and robotics, and augmented reality and virtual reality—could be combined and applied to a project with real world context.

Digital fabrication is one of the emerging technologies poised to revolutionise architecture, engineering and construction. We are working with the University of New South Wales to explore how this, alongside other big trends, will change the way we design, manufacture, and even interact with buildings.

Working with our research team, Hank and Alessandra put together a three-semester long investigation where undergraduate students used emerging technology to design, develop, document, and manufacture a 1:1 prototype of a responsive room.

The result is the Centaur Pod—currently on track to be completed in June. The origami-style meeting room, design to expand and contract in response to how it is being used, is inspired by interactive architecture and soft robotics.

Soft robotics is an emerging field of robotics where highly compliant materials like rubber or silicon are used instead of mechanical parts. The result is a safer, more flexible robot—which generally makes it less dangerous for humans to interact with.

“The interesting thing about soft robotics is that you can scale them up quite big,” says Hank. “You can transform large areas with the same principles as smaller areas.” This makes soft robotics particularly attractive for those designing at the building scale.

In the absence of mechanical rods and motors, soft robots are generally activated by pressure differentials. This makes them move in a similar way to invertebrates. Think of a robot with a goo-filled tentacle like an octopus rather than a claw-like mechanical gripper. Who would you rather give you a hug?

This video, by nature, demonstrates some of the possibilities soft robotics opens up. Another great example is Furl, a soft pneumatic pavilion by Interactive Architecture Lab, in which air is used to animate a series of soft silicone cast 'air muscles.'

This is where Alessandra’s passion for biomimicry comes in. Biomimicry is the practice of designing systems based on biological processes. The Centaur Pod is powered by pneumatic air, which allows its origami inspired folds to expand and contract like human muscles. When a person in the Centaur Pod moves, the Pod will sense their movement and adjust its form accordingly. Over time, the Centaur Pod will learn from these interactions and eventually begin to adapt proactively.

“The responsive nature of the space could be fed into activity-based work environments,” says Hank. The Centaur Pod will initially serve as a meeting room—an office building is generally quite controlled, making it a safe place to test its functionality—but Hank and Alessandra hope the Centaur Pod serves as a test-bed for how responsive architecture can be used elsewhere in the built environment.

“Houses may not be a possibility from a psychological perspective,” admits Alessandra. “Humans rely on stability and a fixed place. But what about facades that can react to solar impact or rainfall to harvest sunlight or collect water?”

A prototype of the Centaur Pod, displayed during a UNSW student presentation at our Sydney office. Photo: Daniel Yu / UNSW
The larger set of prototypes developed through the project. Centaur Pod in the back. Photo: Daniel Yu / UNSW
Another prototype showing the human scale. Photo: Daniel Yu / UNSW
And another, showing how each is scaled up. Photo: Daniel Yu / UNSW

Alessandra believes that more responsive buildings are one way we can respond to the global challenge of climate change. For this reason, she proposed another dynamic building concept to Arup’s Global Research Challenge earlier this year.

Every year, our research program makes an open call for projects which take a multi-disciplinary approach to shaping a better world. One of this year’s themes, Artificial Intelligence, called for examples of how we might use complex adaptive systems to overcome particularly wicked challenges. Just as invertebrates aren’t the only organisms which we can model movement on, there are many ways we can incorporate dynamic and responsive movement into our structures.

Once again Alessandra turned to biomimicry for inspiration.

“We went to nature and asked: how can a structure be thinking and be reactive without being controlled by a centralised system?” says Alessandra. “There are two systems we thought about. One is the octopus, which doesn’t have a centralised nervous system. The other is swam intelligence. Swarms of animals such as ants or birds or fish.”

Alessandra wanted to see how swarm of robots might cooperate to make up a larger structure. This macro structure will then be able to change shape and evolve based on the coordinated movement of the smaller robots, which sense and respond to stimuli in their external environment.

“For the built environment we could see this being useful for creating structures that can react to extreme conditions such as earthquakes. It could provide a protective shield for people that are under ruins,” says Alessandra. Her model of swarm intelligence is especially practical for disaster relief because unlike in a centralised system there’s no risk of a main unit going down and the whole system collapsing.

That said, Alessandra’s swarm of robots could also be useful for more pedestrian tasks. One example is creating reusable formwork for concrete. But first, Alessandra has the technical challenge of modelling a system capable of collective decision making. Her initial prototype will consist of three robots that react together to create a larger form which is structurally sound.

“Each individual needs to understand the present condition as well as predict the future condition to understand where to move and what to do,” she says. “It’s very energy consuming to program. But once you achieve this, it’s something incomparable.”

Kilobots are one system for testing robot swarm behaviour, developed by faculty member at Harvard's Wyss institute and can be seen in the video below.

To understand the value of decentralised thinking, one can also look closer to home. Many modern human networks also operate using similar principles. When we work on a building or a major piece of infrastructure, the output is a result of the collective knowledge of our engineers, architectures, planner and designers. The same can be said about the university environment, where architects, roboticists, software engineering and computational designers can collaborate to solve the same problem, as they are here.

“Optimisation comes from many different minds working together,” says Hank. “You really can bring different people from different areas together on one project where they can bring their insights and apply their knowledge on one particular topic.”

Hank and Alessandra hope the outputs of these projects will have similar knock-on effects.

“With all the research we do in computational design we are less interested in using computation for provocative form finding—having a fancy bobbly form just for the sake of it,” says Hank. “We are really interested in bringing computational design from promise to practice—so really working on real projects and real concerns industry has.”

David Madden is an engineer with our buildings group and an avid researcher. He is leading Arup's research contribution to both projects.

Front cover: Alessandra Fabbri, lecturer and research lead of pavilion and robotic swarm projects.


  • Artificial intelligence, machine learning, digital fabrication, robotics, augmented reality and virtual reality will have a profound influence on the future of the built environment. Projects like these allow us to test these ideas, develop them, evaluate them and eventually bring them into practice.
  • The Swarm project has only just been kicked off, so stay tuned for our prototypes and findings.

This story was written by Jeff McAllister, as part of the Research Review series. The series is produced by the Arup Australasia Research team; Alex Sinickas, Bree Trevena and Jeff McAllister with contributions from Sheda and Noel Smyth.

Lead Arup Researcher

David Madden
David is a structural engineer in our Sydney team.

Ask David about:

  • Architectural, computational and structural design at Arup.
  • Working with university researchers and students to prototype designs for industry problems.
  • Responsive and dynamic design.


M Hank Haeusler
Associate Professor M. Hank Haeusler Dipl.-Ing. (Fh) / PhD (SIAL/RMIT) is the Discipline Director of the Bachelor of Computational Design (CoDe) at the Australian School of Architecture + Design (Built Environment) at the University of NSW, Sydney. Since December 2017 Haeusler is also a Professor at the Visual Art Innovation Institute at Central Academy of Fine Arts, Beijing. His research interests include media architecture, digital technology, computational design, interaction design, and ubiquitous computing.

Research TEAM

Alessandra is an Associate Lecturer in Computational Design at UNSW, and is leading the Robotic Swarms and pavilion projects.

Have a problem or a project?

We work with industry partners, governments, universities, startups and community organisations. We do this through research partnerships, and as consultants and facilitators for foresight, research, storytelling and technical writing workshops.

research with us