Geometric reasoning for optimizing backward growing-based feature recognition

Feature recognition has been a weak link in most implementations of integrated CAD/CAM systems for machining. A method called Backward Growing was recently suggested as a simple, but effective means to recognize machining features from the boundary representation (brep) of a designed part. The method promises, through its simplicity, to be applicable to many varieties of machined shapes. However, it has some drawbacks which create difficulties in making intelligent decisions about machining of shapes. In this work, we deal with two such difficulties, and propose a methodology of solving them. We also present some of the work done in implementing an enhanced version of backward growing as a functional part of an open-architecture system for process planning called OSCAP.