Efficient Automated Geometric Feature Recognition through Feature Coding

Abstract

Automated geometric feature recognition (GFR) is a commonly encountered task in the creation of any process planning or design for manufacturing software. This paper describes a new method based on feature coding for automated GFR. An enhanced winged edge data structure including surface type labels and a Multi-Attributed Adjacency Matrix (MAAM) is generated from the CAD model of the given object. The MAAM fully captures the topology and coarse geometry of the object for the purposes of GFR. A simple algorithmic method extracts each feature from the object-MAAM. The feature-MAAM is then processed to generate a unique code which is recognised and interpreted by matching it with entries in a Feature Database. The method is significantly superior to previous GFR methods in terms of computational efficiency and the reduced need to invoke expert rules. Unlike previous methods, the system can handle objects with plane, cylindrical as well as other analytically definable curved faces and can recognise both simple and complex features i.e. those formed by interactions amongst simple features.