An algorithmic design grammar for problem solving

Abstract

Shape grammars, in conjunction with description grammars, have potential for applications beyond stylistic design generation, such as problem solving. However, few case studies elucidate how to map design logic or knowledge onto the development and organization of rules. As such, the application process usually remains manual or controlled by an optimization algorithm. The aim of this study is to create a design grammar which can rely on logic embedded within or between rules to guide the derivation process towards an objective in an automated way. We propose a general approach to develop an algorithmic design grammar which allows logic to be embedded at varying complexity level. To improve the clarity of logic and automate the application of grammars, rules are structured in algorithmic patterns using sequence, selection and iteration. A layout generation problem is used as a case study for demonstration, and design grammars with different deduction strategies are compared regarding their applicability and limitations.