>Hi!
>
>I'm ending my studies in computer science. My final work is the development of
>an expert system in the field of Intelligent CAD. Particularly, the system
>should be able to design the interior of some architectural environments, such
>as an office, starting from some object libraries of furnitures.
>
I think the biggest problem you're going to face is that interior design
involves the visualization and management of space--what one gets must
after all allow for appropriate use. If in addition, one wants the designs
generated to be pleasant, one is going to have to incorporate some
aesthetic constraints in the software.
Ray Lauzzana follows up:
>
>Have you looked at "shape grammar" systems? Generally speaking, they are
>not interactive. But, they do generate design alternatives from basic
>rules given some intitial constraints. Just exactly how this "guidance"
>of the system would work does not seem to be very clear. There are lots
>of models for it, ie. from learning(example-counterexample).
>
Hmmmm. That jogged my memory. L-systems (see *The Algorithmic Beauty of
Plants*, Prusinciewiscz (spelling?) and Lindemayer) might be useful. It's
an interesting book in it's own right; amazing what lifelike plant forms
one can get from very simple algorithms.
Randolph Fritz
randolph@xxxxxxxxxx
>
>I'm ending my studies in computer science. My final work is the development of
>an expert system in the field of Intelligent CAD. Particularly, the system
>should be able to design the interior of some architectural environments, such
>as an office, starting from some object libraries of furnitures.
>
I think the biggest problem you're going to face is that interior design
involves the visualization and management of space--what one gets must
after all allow for appropriate use. If in addition, one wants the designs
generated to be pleasant, one is going to have to incorporate some
aesthetic constraints in the software.
Ray Lauzzana follows up:
>
>Have you looked at "shape grammar" systems? Generally speaking, they are
>not interactive. But, they do generate design alternatives from basic
>rules given some intitial constraints. Just exactly how this "guidance"
>of the system would work does not seem to be very clear. There are lots
>of models for it, ie. from learning(example-counterexample).
>
Hmmmm. That jogged my memory. L-systems (see *The Algorithmic Beauty of
Plants*, Prusinciewiscz (spelling?) and Lindemayer) might be useful. It's
an interesting book in it's own right; amazing what lifelike plant forms
one can get from very simple algorithms.
Randolph Fritz
randolph@xxxxxxxxxx