What automated design is going to do to 3D printing and product customisation

3D printing has delivered the capability to create solid objects from thin air, but it still requires significant input on behalf of the designer to create an original design.

Now the technology of generative design promises to take away some of the human grunt work involved in 3D, and if successful, might herald a new era of mass-customisation of physical objects.

Generative design starts with a description of the problem to be solved, such as the need for a strong but lightweight bulkhead in an aircraft, accompanied by desired qualities or constraints. The software can then determine the optimal design to meet those requirements based on its understanding of the materials to be used.

“It is kind of a new way of designing,” says Autodesk senior designer, Arthur Harsuvanakit. “The key difference is designers no longer have to spend all their time translating their ideas into digital models, which is the grunt work of design roles. And the solutions that the software provides are already tested for structural performance.

“Now it is the designers job to navigate those solutions, and then determine which is the most appropriate.”

The concept is already embedded into two of Autodesk’s products, and relies on cloud computing to handle the complexity of the calculations involved. Harsuvanakit says it is currently used when designing high performance structures such as race cars or aerospace parts, where weight performance is crucial.

However, he suspects it might not be long before the technique is used to created products with broader appeal, such as lightweight bicycles or backpacks.

“It can touch a lot of products where the cost of materials is really important,” Harsuvanakit says. “The software inherently reduces volume while maintaining the structural integrity that you are required to have. So essentially you are accomplishing your goal with less material.

“And we are looking at where generative design can be used to compute very different layouts of floor planning, or wherever something needs to be optimised where we have a lot of data to back that up.”

By keeping the initial constraints loose, the software can generate designs that the designer might have never otherwise conceived.

“That way you are thinking outside of your own comfort zone in terms of form and function,” Harsuvanakit says.

The technology could also enable greater consumer input into the design of objects, such as furniture, as it can easily handle significant variations in design, such as changing a three-legged stool to be four-legged.

“The platform lends itself to having the user of the product participate more and describe their unique design problem or space set up, where the introduction of a new factor isn’t that hard to accommodate,” Harsuvanakit says.

“Traditionally if you asked for thee legs or four legs that is a whole different design process. But changing three legs to four legs means just deleing one component in the set up and kicking of the process again.

Despite its power, the variations in how a problem might be described means we are not yet at the point where physical objects can be conjured into existence, Harry Potter-style, simply through the spoken word.

“We haven’t yet got to the point where we can use natural language and just talk to the software, but we are at the point where we are describing through the software what obstacles the solution needs to avoid, or how it must perform structurally,” Harsuvanakit says.

He says this should give comfort to those designers who fear yet another erosion of their role.

“Once we describe how it is taking the heavy lifting off 3D modelling and putting it more towards describing a discreet problem that satisfies the requirements of your design space, then it gets interesting,” Harsuvanakit says. “It is about how we look at design in a holist view from idea to the manufacturing and assembly, right from the start.”