The IV th Generation of The INDUCTION DESIGN
      Program of Flow   2003  


      applied to  Tsukuba Express / Kashiwanoha-Campus Station    2004/5

 

 

ALGORITHMIC DESIGN INDUCTION DESIGN CITIES program of Flow Wave Motion MAKOTO SEI WATANABE

 

ALGORITHMIC DESIGN INDUCTION DESIGN CITIES program of Flow Wave Motion MAKOTO SEI WATANABE

ALGORITHMIC DESIGN INDUCTION DESIGN CITIES program of Flow Wave Motion MAKOTO SEI WATANABE

 

ALGORITHMIC DESIGN INDUCTION DESIGN CITIES program of Flow Wave Motion MAKOTO SEI WATANABE

 

ALGORITHMIC DESIGN INDUCTION DESIGN CITIES program of Flow Wave Motion MAKOTO SEI WATANABEmovie: image
program of FLOW
600k      0:26 
480x270
  


Fluid City: The program of flow

 

This is an attempt to further develop regions explored in Web Frame.

The designer makes several sketches, grades them himself, and hands them to the program.
The program speculates about the designers intent from the grades, and shows a proposal that it believes will be graded higher.
The designer grades the program's proposal and returns it to the program.
This process is repeated until (theoretically), the designer obtains the wanted "good design".
This project is especially about the process, with the goal of obtaining a better program.

The program consists of an evaluation module and
a generation module.
A neural network drives the evaluation module, and genetic algorithms drive the generation module.
The two parts are combined for automatic generation.

The neural network does bit mapped (analog data) pattern matching and evaluation of graphic features (digital data).

Artificial intelligence is used to narrow down the number of possibilities to pursue.

Using a neural network, genetic algorithms and artificial intelligence, it is quite a luxurious program.

The object of this program is
TSUKUBA EXPRESS (:TX) / Kashiwanoha-Campus Station.
This architecture has a form like the flow of wind and water. I wanted to realize the flow through a method that did not rely on imitating nature or on human caprice. I could have used a fluid dynamics application to generate the form, but that would be borrowing a pattern from nature. What I am looking for here is not imitation.

I also rejected the conventional method of simply drawing it by hand.

In terms of the actual design, several limitation arise from the fact that it will be manufactured in panels, with requirements for coordinating the connections between parts and limiting the number of forms.

It would be very difficult to use conventional drawing by hand to obtain forms that simultaneously solve the limitations, do not draw attention to the limitations, and feel good.