Graphics Reference
In-Depth Information
language like C++ into machine code optimized for performance on speciic
computer hardware. The machine code is then executed by a physical proces-
sor. Similarly, our matter compiler would translate high level requirement into
a design ile that takes advantage of the latest materials and features of your
3D printer. The printer would then build the object described by the design
ile that the matter compiler generated.
Perhaps a wall bracket isn't the best example since it's a simple design chal-
lenge. The power of a matter compiler attached to a 3D printer would become
apparent if you were trying to design and make a complicated machine. If
20 years from now I was invited to design a robot for the next Mars mission,
here's how I'd like to do it.
First, I would describe the unique conditions on Mars. Then I would give
some overall dimensions, for example, a weight range, what raw materials
will be used, and so on. Finally, I would provide the performance speciica-
tions: its speed, its stability, its eficiency, and so on. Then I'd hit “Enter” and
wait for the matter compiler to generate some new design ideas. Someday,
when matter compilers become an everyday design tool, the time-consuming
process of trial and error that characterizes complex design projects will be
handled very quickly.
Interactive evolution—breeding designs
As intelligent as it would be, even a matter compiler wouldn't be able to read
minds. True, the compiler could devise great design solutions based on what
you tell it you need. However, it could not brainstorm with you. What if you and
your design tools could converse and together work through several different
iterations of a design-in-progress?
If a computer could quickly toss out design ideas in rapid iterations, you could
pick the ones you like. Then, the computer could “study” the design suggestions
you chose and make some rapid adjustments to them and offer them back to you.
You could again select your favorite options, return them to the computer, and
again, it would make adjustments and offer you the result yet one more time.
An iterative design process in which a human and a computer bounce ideas
back and forth is called interactive “evolutionary design.” Like biological evolu-
tion but much faster, a computer can reconigure a design using mathematical
algorithms. The beauty of interactive evolution is that the user does not need
to know anything about the inner workings of the computer's process. A user
does not need to know how to use design software or how to calculate optimal
