Design as parameterization: brute-forcing the manufacturing/design problem-space


The singularity is approaching. But computers still aren’t very good at recognizing beauty.

Is it just me or does that chair (it IS a chair, no?) looked really tippy?

Sounds like a job for a genetic algorithm.

Design parameters as fitness function.

My thoughts exactly. Why do an ugly bruteforce when we already have numerous genetic algorithms such as ant-colony optimization and simulated annealing. They already come up with great optimizations nobody thought to try, and I would expect they’re much cheaper to run than all-out bruteforce.

Only problem is designing the fitness function to give you something that you want, and being sure to cull off the absurdities.

If I remember correctly, genetic algorithms are wonderful at finding both local and global minima and maxima in systems that aren’t obvious from the human’s perspective.

“an optimization system finds the best designs, returns them to the starting point, and refines them repeatedly”

Sounds like they are doing, effectively, a genetic algorithm and just not calling it that.

And they’re only automating the uninteresting parts of design, leaving designers more time to do more important things :smile:. And maybe if it makes good design cheaper, we’ll see more of it? crosses fingers

Reminds me of the stuff at the bottom of the (nuclear bomb produced) caves in the short story, Taklamakan by Bruce Sterling. (text) Only instead of happening inside the computer these genetic algorithms were actual physical things, “Autonomous self-assembly proteinaceous biotech. Strictly forbidden by the Nonproliferation Protocols of the Manila Accords of 2037”. This chair, and a thousand slightly different variations tested for fitness, is definitely something the creatures/tech in the story would have produced. And then they would have crawled out of there and eaten you right up!

While its certainly an interesting idea, it’s probably more useful for giving the designer unusual or counter-intuitive ideas on how to solve the design problem, rather than offering useful designs as a final product. Especially if the item in the photo is a representative sample of a ‘good’ or human-selected outcome. Design for manufactureability seems to be limited to things you can produce with a 3D printer, rather than more traditional production methods. Which isn’t necessarily a problem for some design fields, but I would want a design you can weld or bolt together for, say bridges, until someone invents a very large, steel-output mobile 3D printer (and by god, that would be something to see!)

1 Like

Hell, I hope it gets invented just so I can see what happens when it gets a jam

1 Like

This topic was automatically closed after 5 days. New replies are no longer allowed.