Environment and Planning B: Urban Analytics and City Science

It is argued here that we must establish a demonstrably sound, comprehensive, rigorously formalized theoretical foundation upon which to base practical computer-aided architectural design software-development efforts, and a general approach to this task is suggested.

First, the basic types of primitives, structures, and operations that we might employ are considered. Next, the concept of a formal architectural language is developed. Finally, design synthesis is viewed as a process of searching within such a language to find a particular design which, under specified algorithms that establish the semantics of the language, has acceptable interpretations.

The implications of a program of formalization of architectural knowledge along these lines for research, development, teaching, and design practice are discussed.

Although substantial literature exists on the properties of formal grammars, much less has been written on the use of grammars for describing the languages which they are capable of explaining. Thus we see a well-developed theory concerning the expressive (generative) power of different kinds of formal grammars and powerful algorithmic methods (analytical) for the languages defined by these grammars. For natural spoken and written languages, grammars have been used for explaining their structure but for the two-dimensional generalization to designs, pictures, images, and fine arts, which has been known since 1964, almost no use has been made of grammars.

The purpose of this paper is to call attention to the powerful dormant tools that can be used in the design arts. Some of the benefits of using these tools for describing existing (natural) design languages are discussed. How such uses may direct us from the more formal design arts into the fine arts is discussed by reporting progress on building a grammar for a class of contemporary paintings.

A design world describes knowledge about possible designs. The characteristics of design worlds in engineering and architecture are compared. An implementation of three-dimensional shape grammars is outlined and the description of composite spatial relations which arise in generated shapes is considered.

Several computer programs that enumerate rectangular dissections as solutions to certain layout problems have established a distinct paradigm for dealing with the crucial theoretical issues involved. In the present paper an extension of the paradigm is suggested, to include ‘loosely packed arrangements of rectangles’, which are of wider applicability in an architectural context. Orthogonal structures are introduced to represent these arrangements and the conditions of well-formedness for these structures are established. A grammar to enumerate orthogonal structures is presented and it is suggested that best use is made of the grammar if it is incorporated into a generative expert system, able to serve as a vehicle to discover, encode, and utilize a broad range of constraints and criteria in the generation of layout alternatives.

This paper is concerned with problems that arise from attempts to integrate the components of a building model in a database that reflects the current status of a design. It is based on work that began in 1976 at the Architectural Research Laboratory of the University of Michigan. This effort, which was the work of many individuals, had as its goal the construction of an integrated computer-aided building design system. Many of these ideas have been implemented in a working experimental modeling system, ARCH:MODEL. From a series of both practical and theoretical experiences using and building this system several difficult questions have come up that need research attention. Among these are: how can a design database be made consistent and how can consistency, both spatial and nonspatial, be maintained as the design develops? These questions are related to the general problem of what is the relationship between the design process and the designed object. It is not the intent of this paper to provide answers to these problems but to argue for their importance and to provide some direction for research.

It is no longer unusual to find automated systems as partners in the architectural design process. Most, if not all, of these systems are limited to solving well-structured problems efficiently and accurately. The use of these systems relies on manual decomposition of complex problems into more limited, well-defined ones and the subsequent recomposition of the solutions into comprehensive ones. In this paper a formal representation is developed for the process of problem structuring to enable decomposition and recomposition of architectural problems. This consists of a heuristic method for manipulating the parameters of a general purpose generate-and-test mechanism which is capable of solving well-defined problems.

Research into a generalized logic modelling system linked to machine interpretation of drawings and text is described. This work is related to a view of design and builds on experience of working integrated design systems. Questions are posed for future computer-aided design systems.

This paper is an attempt to track the path of architectural design both on design-theoretical and on practical (computer-aided design) planes. It begins with the practical problems of design which can be couched in terms of mathematical programming problems. The potential for applications of artificial intelligence and knowledge-based systems in architectural design is then considered. Last the place of architectural design within design theory is discussed.