Creating Polytope Representations of Design Spaces for Visual Exploration Using Consistency Techniques

Abstract

Understanding the limits of a design is an important aspect of the design process. When mathematical models are constructed to describe a design concept, the limits are typically expressed as constraints involving the variables of that concept. The set of values for the design variables that do not violate constraints constitute the design space of that concept. In this work, we transform a parametric design problem into a geometry problem thereby enabling computational geometry algorithms to support design exploration. A polytope-based representation is presented to geometrically approximate the design space. The design space is represented as a finite set of (at most) three-dimensional (possibly nonconvex) polytopes, i.e., points, intervals, polygons, and polyhedra. The algorithm for constructing the design space is developed by interpreting constraint-consistency algorithms as computational-geometric operations and consequently extending (3,2)-consistency algorithm for polytope representations. A simple example of a fingernail clipper design is used to illustrate the approach.