Evolving Digital Morphogenesis by Means of Biology and Computer Science

Daniel Davis – 12 November 2009

In my undergraduate thesis (BArch Victoria University, Wellington) I developed a novel method for designing with genetic algorithms. Rather than using a genetic algorithm to seek a single ‘optimal’ design, I invented a method for identifying a range of Pareto-optimal designs. Using Java I created an application that evolved a population of design outcomes using a genetic algorithm. These outcomes were sorted into multi-dimensional Pareto-fronts that the designer could then explore using a set of sliders (video). Using this application, the designer and the computer worked synergistically to explore potentials in the design space of genetic algorithms.

I applied my Pareto-optimal genetic algorithm design method to the conversion of a warehouse in Wellington, New Zealand. Working at a range of scales, I demonstrated the method being applied to space planning, structural analysis, and the design of various stairs and chairs. My thesis was selected as one of four projects from Victoria University to attend the New Zealand Institute of Architects student design awards.

Every panel together
The five presentation panels

 Panel One

davis_undergrad_1

 Panel Two

A floor plan selected for homogeneity
A floor plan selected for homogeneity
A floor plan selected for variation
A floor plan selected for variation
A floor plan selected double height spaces
A floor plan selected double height spaces

Panel Three

Structure selected for its constructability
Structure selected for its constructability
Structure selected to minimise window obstruction
Structure selected to minimise window obstruction
Structure selected for its regularity
Structure selected for its regularity

Pareto-optimal Structural Design Process

Panel Four

Stair One
Stair One
Stair Two
Stair Two
Stair Three
Stair Three

Panel Five

davis_undergrad_5
panel5_chair_plastic
panel5_chair_wood
panel5_chair_steel