You may have heard me going on at length about my genetic algorithm project. Here’s a practical application by Gennaro Senatore, in which he programs my GA framework to evolve structures that have recognizably architectural features.
The best thing is that Gennaro’s thesis is largely the paper I would have written myself – now I don’t have to!
Abstract: The thesis discusses the possibility to build a design methodology based on the utilization of generative rule systems, evolutionary techniques and performance evaluation tools. This is conducted through the implementation of a computational system in which spatial configurations can be defined by a set of instructions framed in hierarchical data structures. These are graphically interpreted to create their geometrical expression which is evaluated according to design criteria such as structural stability and accessibility of its basic components. String rewriting Lindenmayer systems and Genetic Programming are combined to first create the definition of the problem and afterwards evolve possible answers to it. The aim is to integrate the generation of forms and the evaluation of their performances in order to design spatial configurations whose morphology emerges out of a process of formation rather than imposed by a fixed and predetermined representation. The adaptation of such rules system under the pressure of encoded design criteria resolves in the emergence of forms which become, hence, the expression of high level of abstractions.