Matias del Campo’s interest in bringing artificial intelligence into the field of architecture started in conversations in 1996 with those at the Artificial Intelligence Institute in Vienna, the town where he was raised. “I saw the possibilities,” says del Campo, an associate professor at Taubman College.
At that time, those possibilities were completely theoretical since there was no large-scale machine learning. More than two decades later, del Campo has spent the past few years researching and developing applications in the fast-advancing field of artificial intelligence that are making significant contributions to the field of architecture. He directs the University of Michigan’s Architecture and Artificial Intelligence Laboratory, which was founded in January 2021 and brings together faculty from robotics and computer science, as well as Ph.D. and graduate students from computer science and architecture.
Taubman College has been at the forefront of AI, del Campo says. The focus has accelerated in the last couple of years as “the computational power and the algorithms really caught up with things that we had in our minds, so we finally could really apply them.”
His interest in architecture began when he was 14 and had a crush on a girl. Her father was an architect and, eager to impress her, del Campo read about his work. Then he thought, “This is exactly what I want to do.” This ignited a passion for architecture that “became an obsession.” He became fascinated with learning how space influences humans. “Good architecture basically touches you in some way,” del Campos says. In an interesting twist, the girl’s father ended up being one of del Campo’s professors as an undergraduate student in Vienna.
In addition to his work at U-M, del Campo is co-founder of an architecture practice called SPAN. He specializes in applying AI tools to architecture in two areas known in the field of engineering as “the wicked and the tame problem.” The “wicked problem” refers to areas that are hard to capture, like aesthetics, sensibility, and the cultural and artistic qualities of an architectural project. The “tame problem” areas include aspects of optimization, structural engineering, and material properties. In both situations, AI can harness large amounts of previously generated data.
“We can dive into our own history as a discipline and extract new information out of existing data in order to inform future designs,” del Campo says.
He explains that when he feeds images into an “artificial neural network,” it will provide similar but not identical images that capture the same quality. “It might not be inventing anything new, but it’s really good in interpolating and reassembling this information into something that is surprising and beautiful and interesting.” As a result, AI can save time and money by speeding up certain mundane processes like verifying whether a plan is up to code. If it’s done automatically, it could save an architect several days of time, del Campo notes.
It can also inform design in different ways, “providing a designer with solutions that might be unexpected and different” than what was initially conceived. He views working with computers and machines as more of a dialogue than just imposing his will on the machine.
He is particularly proud of the Robot Garden that he began working on four years ago, a collaboration with U-M’s Department of Computer Science and Michigan Robotics. It allowed del Campo to, for the first time, translate elements he had been trying to design with machine learning into a physical construction project. The goal was to create a garden to test the agility of bipedal robots on various surfaces, such as gravel, rocks, and sand. This has real-world applications, like when robots deliver packages. “It’s more challenging than you would think,” del Campo says.
The garden, located next to the Ford Robotics Building on U-M’s North Campus, opened in September 2021. The project is important, del Campo notes, since it proved that the ideas being developed on how to apply AI and architecture are not only feasible but actually make things faster and easier. Still, the adoption will be limited if contractors and construction don’t adopt these new methods. “We as architects can show possibilities, but the industry must have the guts to bring them up to scale,” he says.
Beyond his practice, del Campo also is at the forefront of incorporating technology into teaching. During the pandemic, he was concerned that a practice that is three dimensional had to rely on two-dimensional surfaces for presentation. So he was in touch with a company in Australia which provided a program called HoloLens, an augmented reality software that allows students to work on a live presentation platform in 3D, so that students can see in space where a piece should go.
Since joining the Taubman College faculty in 2014, del Campo has been most impressed by the support for interdisciplinary work that he’s able to pursue: “I’ve not seen that anywhere else in depth on that scale.” And he says Taubman College has been extremely supportive of AI: “They’re not afraid to try new things.”
Some of his projects were postponed due to COVID. He’s excited about one that is resuming: a kindergarten in China. His relationship with that country began with a design of the Austrian Pavilion at the 2010 Shanghai World Expo. He’s also working on a project called The Common House, which uses AI to generate apartment plans for all over the world that are optimized for certain conditions like climate, ecology, and the economy. This offers the possibility of architects being able to successfully work in many other countries. He just finished a book focused on AI and architecture, called Neural Architecture: Design and AI, which is due out by February 2022.
Now that those possibilities that del Campo noted in 1996 are becoming reality, he is excited by the way that others are adopting the methods that he developed “and making them their own.” AI, he says, “is going to happen quite quickly. A lot of people will pick it up because it’s just practical. It’s going to make things easier for a lot of people.”