Computer-aided design systems for the 21st century: some design guidelines

This paper proposes nine design principles for a new generation of computer-aided design (CAD) systems that actively support the decision making and problem solving activities of environmental design. Foremost among these are: a meaningful internal object-based representation of the artifact being designed within its environmental context, a collaborative problem solving paradigm in which the human designer and the computer form a complementary partnership, and, the notion of decision-support tools rather than predefined solutions. Two prototype computer-aided design systems implemented by the CAD Research Center that embody most of these concepts are described. ICADS (Intelligent Computer-Aided Design System) incorporates multiple expert agents in domains such as natural and artificial lighting, noise control, structural system selection, climatic determinants, and energy conservation. Given a particular building design context, the agents in ICADS draw upon their own expertise and several knowledgebases as they monitor the actions of the human designer and collaborate opportunistically. KOALA (Knowledge-Based Object-Agent Collaboration) builds on the multi-agent concepts embodied in ICADS by the addition of two kinds of agents. Mentor agents represent the interests of selected objects within the ontology of the design environment. In the implemented KOALA system building spaces are represented by agents capable of collaborating with each other, with domain agents for the provision of expert services, and with the human designer. Facilitator agents listen in on the communications among mentor agents to detect conflicts and moderate arguments. While both of these prototype systems are limited in scope by focussing on the earliest design stages and restricted in their understanding of the inherent complexity of a design state, they nevertheless promise a paradigm shift in computer-aided design