Autodesk’s New Toronto Office Displays Algorithm-Driven Generative Design

Autodesk recently unveiled its new Toronto space inside MaRS Discovery District—but it wasn’t a typical ribbon cutting. It marked the first large-scale generatively designed office ever constructed.

Spanning three floors and 60,000 square feet, the American multinational company chose Toronto to display the architectural power of a new tool Autodesk has incubated over several years. It’s called Project Discovery, a generative design tool.

“We automated the design of an office space. But the result is that it augments the skills of a designer,” said Gordon Kurtenbach, Autodesk’s head of research.

Project Discovery combines human ingenuity and machine learning, using intelligent algorithms to factor in engineering constraints to create thousands of design possibilities. These constraints are elements that can’t be moved or adjusted including windows, stairs, elevators and the physical floor space. Generative design takes these parameters into account and pulls in the individual preferences of the people that work in the space to humanize the design.

Kurtenbach said they met with teams and surveyed more than 250 employees to learn what matters to them within an office. For Autodesk Toronto, there were six measurable preferences, including distance to neighbours and amenities, availability to daylight, visual distractions and views to outside.

Autodesk Toronto took the personal input, quantified the data and then synthesized that into the algorithm that generated office layouts. In the first meeting with the research team running the project, Kurtenbach recalled being showed what he thought were eight designs.

“But those were just the eight families of designs… In the end, we generated 10,000 possibilities and selected the best from that,” he said.

Each team works within what Autodesk calls a neighbourhood, one of 15 spaces the teams had the chance to customize with murals and more.
“When you have a mix of teams with different work styles, a big question is: what does a productive office look like?”

Kurtenbach assured that architects are still needed in the design and delivery of a space, but that Project Discovery can enhance what they do. The software represents a co-design between human and computer that makes data-informed trade-offs to create an optimized office.

“For the first time, we can fulfill the promise of design: to understand the intention and understand the connection of what you commit to before you actually put it on the planet,” said Jeff Kowalski, Autodesk’s chief technology officer.

The software allows stakeholders to visualize how design elements are interconnected and how small preferences can shift the design.

“If we have limited resources in form of energy, labour, natural materials, and we’re seeking to optimize them on the planet, the way that all architecture has been performed to date is that we have shipped out betas. We understand it after it is built and then wish we could have built something else,” he added.

Each design changes the boundaries between the teams and rearranges desks to meet the complex goals, helping an architect navigate through those options.

“We are combining human intelligence and machine intelligence… to create better outcomes,” said Kowalski.

While Autodesk was the first customer to use their own generative design software to conceptualize office layouts, Kurtenbach said that other companies are beginning to use the tool for similar purposes.

Similarly named generative design tool Project Dreamcatcher has been used for consumer products. Under Armour used the software to consider weight, style, material and cost to create the first consumer-grade 3D printed performance trainer.

Autodesk will be at Toronto’s next Techfest on Oct. 26.