Complexity measures as a basis for mass customization of novel designs

In this paper we present a computational approach to the integration of mass customization into the design of novel solutions. We propose the use of entropy as a function of the diversity of solutions that can be generated within a design space. This work demonstrates that the potential of design systems to generate novel solutions can be estimated using complexity measures. This principle is implemented in an evolutionary system for the design of automotive instrument panels that display situation-relevant information in configurations that adapt to traffic conditions and driving actions. This sample application shows that the application of complexity maximization as a selection criterion in evolutionary design systems yields a large variety of solutions of high fitness. We also present guidelines for future developments.