The process of building design is a means of meeting client expectations for a building, followed by a set of comprehensive standards and codes relating to the design, construction, and maintenance of buildings. The rapidly rising volume of data along with increasing client expectations inspired many researchers to develop new computerized techniques to automate building design process. These techniques can be broken down into two broad categories: automated computer-aided design tools (e.g., development of building information modeling or BIM) and measuring or modeling client expectations (e.g., use of facial expressions to find psychological expectations of the client). Advances in artificial intelligence (AI) and machine learning have made possible new approaches to the automation of design. This represents a shift from a focus on client-designer or designer-technology relationships to a more intelligent and independent client-technology communication. In this paper, a novel AI system called intelligent designer is proposed to understand (or learn) the client's need and expectations and generate valid designs. The design environment (i.e., the interaction between the client and the design) is formulated as a Markov decision process (MDP) and a mathematical framework is provided for making design decisions in situations where new designs are partly random (as they are influenced by the client's feedback) and partly under the control of the computer (as they are influenced by the regulations, standards, and guidelines). The approach is demonstrated using a window design experiment.
All Science Journal Classification (ASJC) codes
- Control and Systems Engineering
- Civil and Structural Engineering
- Building and Construction