14 February 2011

Generative Desgin for the physical world

Recently I started looking at 3D from different points of view and got a taste for the Physical World. While it's relatively easy to make abstract models that will live in a virtual world, the same does not apply for the this world.

I went to a pretty interesting workshop recently: Generative Design for Robotic Manufacture.The goal was to learn how to take paper folded prototypes into Rhino, parameterize them using Grasshopper, do verlet physics simulations using Kangaroo and hopefully use Lobster IK to simulate the robot movements too. This workflow would help tweak prototypes before they could finally be produced by RoboFold robots.

Grasshopper surprised me in a nice way, with an intuitive visual programming interface similar to Xpresso or better yet, MaxMSP.

I heard "parametrize" this and "parametric" that without knowing for sure. Well, it turns out that "parameter" is architect lingo for "variable" with was comforting.

The workshop kicked off with making paper prototypes. The goal was to play with different techniques(like repetitions, intersections,etc.) to get a feel for what can what can't be done and what kind of folds restrict movement and how. It was a very interesting process. The computer might make some aspects easier, but nothing beats the world world (so far). I am easily fooled that the computer is much more than a tool, using it on a daily basis, but I was pleasantly surprised by the process of using indispensable simple design tools. I didn't need any manual to use them(pen,paper,ruler and scalpel.), I could dive right an quickly prototype.

The next step was to scan a design and get it into Rhino, from there we exported models to .dxf so we can also try them in Rigid Origami. The process sounded easy, just marking mountain and valley folds in red and blue lines for Rigid Origami, but in practice I found that simpler designs worked better than complex ones which did not work at all. You can see a screen grab from Rhino showing the same fold pattern in 3 version, one simpler than the other. Also, I've added a couple of images from Rigid Origami. The next step was to create a parametric version in Grashpopper, which is why the surface was ruled mostly using even steps. After that we played with Kangaroo to get a feel for doing physical simulations on the model.

scanned surface drawn in Rhino

scanned surface imported in Rigid Origami

The second day we got paired to design surfaces that would fit a pretty nifty window pattern. I got paired with Dr Kontovourkis Odysseas who teaches Architecture at the University of Nicosia in Cyprus. I must admit I was a bit overwhelmed at the workshop, not only by the amount of new information (which made me wish I picked up Rhino/Grasshopper earlier), but also by the rest of the attendees. I was the youngest, and came from a pretty different background, where the closest thing would be making 3D models in an abstract world. Most of the attendees were architects from different parts of Europe. Most of them were German speakers, although they weren't from Germany. My colleague was kind enough to get me up to speed with Grasshopper and we started using designs from the previous day to fit on the window pattern.

Again simplicity won. Initially we had a folding with quite a few folds in, but it turned out not only it was hard to parameterize, but it also wasn't very scalable. We started from scratch and got a parametric design.

parameterized design in Grasshopper

It was a challenging experience, but a fulfilling one. The talented people at Metropoltan Works were a pleasure to learn from and I was impressed with their facilities for 3D scanning and printing.

I seem to find cutting and creasing paper very relaxing and I enjoy the different type of thought process. Looks like I found a way to do 3D, but spend a bit more time away from the computer.

No comments: