Abstract
Feature-based modelling is an approach being studied for the next generation of intelligent CAD environments. In particular, feature mapping, i.e. how to support automatically the different ways specialists view the same object, has been identified as an important element in feature-based design systems. Consequently, feature representation and reasoning tools based on the representation are crucial issues. This paper addresses the problems of representing features in a form more suitable for feature transformation and the related feature reasoning mechanism. Our approach combines methodologies from both design and features and feature recognition. Geometric constraints are used within symbolic definitions of features and also to define their relative positions and orientations. A graph grammar parsing is combined with knowledge-based techniques to derive feature information and propagate constraints. The proposed approach can be used for the transformation of feature information and to facilitate feature mapping