Knowledge-Based Creation of an Architectural 3-D Model from 2-D Drawings

Drawings have been the primary means of communication in architectural design for centuries. Architectural drawings of a building are typically two dimensional orthogonal projections except for perspective and axonometric views. Each drawing alone does not represent the whole building; however, a collection of these drawings includes all the details necessary to describe a building.

3-D object-oriented design systems have been gaining popularity in the last few years. A 3-D object system requires features, geometry, relationships and attributes of 3-D objects. This information must be accessed quickly and efficiently in order to fully take advantage of a 3-D object system. However, there is no system today that is able to interpret multiple 2-D architectural drawings and to automatically create a 3-D building model. Presently, designers manually transform the drawings into a 3-D model which is a time-consuming and inefficient exercise. This paper describes a methodology for developing such a system.

To study the issues involved in the interpretation process, we implemented a prototype knowledge-based system called AutoMOD (Automatic 3-D Modeler). The prototype system utilizes a number of existing tools, including Nexpert Object (for object representation and reasoning), AutoCAD (for making input drawings and displaying the output 3-D model), THINKC (for building the interface between Nexpert Object and AutoCAD) and HyperCard

The AutoMOD system has been demonstrated successfully for interpreting simplified 2-D architectural drawings and for building a 3-D object model based on interpretations of the 2-D drawings.