An exploration into evolutionary models for non-routine design

Non-routine design is characterized by the lack of knowledge regarding the relationships between the given requirements and the form required to satisfy those requirements. Thus a knowledge-lean methodology as characterized by evolutionary models seems suitable for non-routine design. However, evolutionary methods to date as characterized by Genetic Algorithms (GAs) have been developed mainly for optimization and machine learning. This paper explores some of the issues involved in the use of evolutionary models as computational models in non-routine design.