Using high-level constraints to aid space planning applications in computer-aided design

This thesis describes a system that aids in solving space planning and layout design problems through an innovative use of constraints. Space planning problem solving is aided by constrained manipulation, a trial-and-error approach, and modelling the scaling properties of objects. Constraints are used to model various physical properties of solid objects (including gravity and friction) and to control the way in which objects scale. The system runs on a Silicon Graphics workstation supporting interactive, real-time, 3-dimensional graphics that aid in visualizing designs. The problems handled by our prototype are restricted to objects representable with rectangular solids oriented along the cartesian axes. This system was used by designers and novice computer users to solve a studio layout problem. Constraints were found to aid in the manipulation of objects by providing predictable behaviour for objects. The test results indicate that subjects learn the behaviour of objects easily and use this information in useful ways. In addition, there are indications that constraints can reduce users'cognitive loads at the user interface level. The coupling of constraints with the interactive, 3-dimensional aspect of the system encourages the use of the system at earlier stages of the design process than simple geometric drafting systems. (Abstract shortened by UMI.)