Designing Biomorphs with an Interactive Genetic Algorithm

We describe an interactive version of the Genetic Algorithm which can be used to evolve Fourier series based \biomorphs." The user plays the role of a binary-valued objective function. The usual Genetic operators of Selection, Mutation, and Crossover are applied normally. We discuss the Interactive Genetic Algorithm's use as a teaching tool and in domains in which measures of utility are di cult or impossible to specify mathematically. We argue that the evolution of biomorphs can be considered an ergonomic design problem, and with the evolution of biomorphs as an example, we argue that a large class of design problems can be cast as \imprecise optimization" and \solved" with Interactive Genetic Algorithms. Thus IGAs can be used for applications that fall somewhere between AI, optimization, and CAD.