INTERACTIVE VISUALISATION SYSTEMS FOR CONCEPTUAL. BUILDING DESIGN: A PRACTICAL APPROACH
M. Y. Rafiq, M. Beck and I. Packham
University of Plymouth, UK
E-mail: mrafiq@plymouth. ac. uk
At the conceptual stage of the design process where only a partial specification for a design is available and due to fuzzy nature of information at this stage it is difficult to program every design requirements. Experience has shown that evolutionary computation EC, (particularly the genetic algorithm) to be an effective decision support tool for conceptual design. To make EC useful in this stage of the design it needs strong human interaction and guidance to lead the search in discrete regions of the search space to explore and discover more appropriate design concepts. Humans are extremely good at perceptual evaluation of designs according to criteria that are extremely hard to program (Eckert et al., 1999). As a result, they can provide useful fitness evaluation for interactive evolutionary systems. They can also include personal preferences to lead the search and exploration to a preferred direction. This kind of interaction is extremely important to satisfy design/client requirements, particularly at the conceptual stage of the design process. This paper introduces a novel approach which demonstrates that interactive use of evolutionary computation, assisted by visualisation tools, leads to a human-led search. A system which support human-led search and it is based on an interactive visualisation clustered genetic algorithm, developed by Packham and coworkers (Packham, 2003; Packham and Denham, 2003; Packham et al., 2004; Rafiq et al., 2004), is introduced and its application on an example of a multi-disciplinary decision making process is demonstrated.
For over a decade not only have techniques from the field of Evolutionary Computation, EC (particularly the Genetic Algorithm, GA), played a major role in design optimisation, but they are also seen as an analogy of design methodologies (Gero et al., 1997). More recently their role has been recognised as support tool for conceptual design (Goldberg, 2002).
The GA is typically a knowledge lean search process, and while this can be seen as one of its major strengths thus providing it with its wide applicability; it can be a hindrance when attempting to incorporate the richness of human knowledge, experience and insight into the search process.
This is particularly exemplified at the conceptual stage of the design process where: Firstly, only a partial specification for a design is available and due to fuzzy nature of information at this stage it is difficult to program every design requirements. It is argued that EC approaches need strong human interaction and guidance to lead the search in discrete regions of the search space to explore and discover more appropriate design concepts; Secondly due to the ill-defined nature of requirements the process itself is not one of search, but rather an exploration of the requirements and the potential solutions to those requirements (Gero, 1993; Maher et al., 1995).
However, with an ancestry in the artistic creations of Dawkins (1986) a new field of Interactive Evolutionary Computing (IEC) has begun to emerge and has been adapted to meet to a wide spectrum of application areas (Takagi, 2001).
M. Pandey et al. (eds), Advances in Engineering Structures, Mechanics & Construction, 595-606. © 2006 Springer. Printed in the Netherlands.
The authors believe that human-led IEC visualisation tools assisted by the search and exploration power of EC could be a powerful tool for human designers. To this end, the next section, discusses and presents a broad categorisation of visualisation and human iteration approaches adopted in EC systems. More detailed examples of visual and interactive systems as applied to conceptual building design will follow. Finally the paper will describe research on the Interactive and Visualisation and Clustering Genetic Algorithm (IVCGA).