| This dissertation demonstrates a new automated design paradigm that merges a shape grammar based representation and generation technique, an evaluation scheme associated with the grammar rules and an insightful optimization algorithm, namely robust activity analysis. In this methodology, a shape grammar for the application domain of interest is created, an evaluation scheme capable of providing feedback on both partial and completed designs is generated, and robust activity analysis is used with the grammar and the evaluation scheme to characterize the design space and determine optimally directed selection of rules and parameters.;Each of the three components of the proposed methodology is demonstrated through examples---the use of shape grammars for knowledge representation and design generation through rule sets for coffeemakers (a consumer product where function drives form) and MEMS resonators (an electro-mechanical device with coupled form and function), the evaluation scheme through a technique for associating manufacturing cost expressions with the rules of the coffeemaker grammar and robust activity analysis through a photoconductor-developer design example. Finally, a system is demonstrated for a bulk manufacturing process design problem illustrating how shape grammars, evaluation schemes and robust activity analysis can be integrated together. Through these examples, the advantages of the proposed design paradigm vis-a-vis the current computer based design approaches are brought forth. |
