Ž.
Automation in Construction 11 2002 1–22
www.elsevier.comrlocaterautcon
Design space navigation in generative design systems
Sheng-Fen Chien
a,)
, Ulrich Flemming
b
a
Department of Architecture, National Taiwan UniÕersity of Science and Technology, Taipei 106, Taiwan
b
()
School of Architecture and Institute for Complex Engineered Systems ICES , Carnegie Mellon UniÕersity, Pittsburgh, PA 15213, USA
Accepted 29 June 2000
Abstract
Generative design systems make it easier for designers to generate and explore design alternatives, but the amount of
information generated during a design session can become very large. Intelligent navigation aids are needed if designers
wish to access the information they generate with ease. We present a comprehensive approach to support information
navigation in requirement-driven generative design systems, which gain their power form explicit representations of design
requirements, which in turn add to the information generated by the system. Our approach takes into account studies dealing
with human spatial cognition, wayfinding in physical environments, and information navigation in electronic media. We
structure the information to be accessed in terms of a five-dimensional design space model that applies across generative
design systems of the type considered here. The model structure supports basic generic navigation operations along its five
dimensions. We validated the model in the context of the SEED-Layout system and used it to extend the built-in navigation
tools of the system through novel ones, which we subjected to a limited usability study. The study suggests that these tools
have promise and warrant further investigation. q 2002 Elsevier Science B.V. All rights reserved.
Keywords: SEED-Layout system; Generative design system; Navigation
1. Introduction
In this article, the term generatiÕe design system
does not refer to a system intended to automate the
design process. Rather, it refers to a system that is
able to take a more active part in the generation of a
computable design description than traditional com-
Ž.
puter-aided design CAD systems are able to do; but
the design process itself may remain largely under
the designer’s control. Work on generative design
systems is motivated by a major bottleneck faced by
sophisticated analysis, simulation and rendering soft-
)
Corresponding author.
ware that needs — as input — a design description
of sufficient detail. Creating such a description or
design model with traditional CAD tools is ex-
tremely tedious, time-consuming and error-prone.
Advances in hardware and computer graphics have
not alleviated this bottleneck; rather, they have made
it more pronounced:
... in any Virtual Reality system, ... many of
the technical tasks are much more difficult when
one uses models of meaningful complexity, con-
sisting of tens of thousands of primitive elements.
Developing and maintaining a model of this size
is an effort comparable to developing and main-
taining a program of such a size, and the software
engineering techniques necessary for big pro-
0926-5805r02r$ - see front matter q2002 Elsevier Science B.V. All rights reserved.
Ž.
PII: S0926-5805 00 00084-4