Intelligent Systems for Supporting Architectural Design

Design can be considered a process leading to the definition of a physical form that achieves a certain predefined set of objectives. The process comprises three distinct operations: (1) definition of the desired set of performance criteria (design goals), (2) production of alternative design solutions, (3) evaluation of the expected performances of alternative design solutions, and comparing them to predefined criteria. Difficulties arise in performing each one of the three operations, as well as in combining them into a purposeful, unified process. First, it is difficult to define the desired performance criteria prior to and independently of, the search for an acceptable solution that achieves them, since many aspects of the desired criteria can only be discovered through the search for an acceptable solution. Furthermore, the search for such a solution may well alter the definition of these criteria, as new criteria and incompatibilities between existing criteria are discovered. Second the generation of a design solution is a task demanding creativity, judgement, and experience, all three of which are difficult to define, teach, and otherwise capture in some explicit manner. Third, it is difficult to evaluate the expected performances of alternative design solutions and to compare them to the predefined criteria. Design parameters interact with each other in complex ways, which cause effects and side effects. Predicting the expected performances of even primary effects involves extrapolating non-physical characteristics from the proposed solution's physical organization, a process which relies on a host of assumptions (physical, sociological, psychological, etc.) and hence is seldom a reliable measure. A fourth problem arises from the need to coordinate the three operations in an iterative process that will converge on an acceptable design solution in reasonable time. Computational techniques that were developed in the past to assist designers in performing the above mentioned activities have shown limitations and proved inadequate to a large degree. In this paper we discuss the work in progress aimed at developing an intelligent support system for building and architectural design, which will be able to play a decisive role in the definition, evaluation and putting into effect of the design choices