Wednesday, March 22, 2017

Architectural 3D Printing Moves Forward with AI

Wow, the field of machine learning and architectural 3D printing is moving fast. I wasn't planning to write about the topic again so soon but I didn't want to let these examples pass without sharing them first. Eureka Magazine has an interesting article up about a small London-based 3D printing firm, Ai Build, which people are going to want to check out. To date, they hadn't been on my radar either. Lately I've been impressed following MX3D's progress on a 3D printed metal bridge in Amsterdam. Ai Build's "Daedalus Pavilion" has similar goals so it's a sign the sector is very competitive. The technology is finally out of the lab but still needs further field research. Recommending architectural 3D printing for anything other than architectural features at this point might be risky, on the other hand, construction company Cazza says it's up to the challenge. They're determined to build a full-scale skyscraper in the Middle East using cementitious 3D printing techniques. The interesting thing here is that while some of the technology is still to be developed for the project, it also piggy backs on well-known and established construction techniques. The tower by Cazza will use cranes as a base of support for the 3D printers. This seems like a really efficient approach in my opinion and is different than the prefabrication techniques, say, used to accelerate Barcelona's Sagrada Familia's completion.

Returning to Ai Build's Daedalus Pavilion, we wish to stress that as this type of technology becomes more widespread, and more and more studios develop the expertise to execute complex generative design programmes, clients are encouraged to raise their expectations as well. This strategy creates a bulwark for our built environment against mediocre 3D printed blobs. Referring to the London installation, descriptively it's a symmetrically deformed surface; maybe some type of hyper-surface. But ultimately these shapes already have a history of use in architecture (in concrete shells). The difference here is that the structure was built by robots with some very interesting underlying software implementations. As people walk around it in a gallery however, I'm not sure how well this story unfolds. It never really breaks away from its nature as an architectural feature. Its engineering challenges are a bit better expressed in my opinion. It looks amazingly delicate. Like a person could lift it up with one arm. Here to keep the whole thing from snapping in half on opening night is the engineering expertise of Arup. I love structural engineering and even love designing strip malls and warehouses, but one has to admit calculating out the forces on lattice structures as shown is not something within every shop's capability to do. Quickly touching on the software before ending, the coders implemented deep learning algorithms during the production phase to improve the accuracy and speed of manufacturing. The approach appears to be quite computationally heavy, hence the involvement by NAVDIA, which I'm assuming brings a substantial amount of electrical engineering and computer science expertise to the project.

A post shared by AI BUILD (@ai_build) on

No comments: