In current commercial feature modeling systems, support for direct manipulation of features is not commonly available. This is partly due to the strong reliance of such systems on constraints, but also to the lack of speed of current constraint solvers. In this paper, an approach to the optimization of geometric constraint solving for direct manipulation of feature dimensions, orientation, and position is described. Details are provided on how this approach was successfully implemented in the Spiff feature modeling system.

In intelligent computer-aided design the concept of intelligence is related to that of integration. Using feature-based computer-aided design models is thought to make a complete integration. This paper presents a feature recognition approach based on the use of a feature grammar. Given the complexity of feature recognition in interactions, the basic idea of the approach is to find the latent and logical structure of features in interaction. The approach includes five main phases. The first phase, called regioning, identifies the potential zones for the birth of features. The second phase, called virtual extension, builds links and virtual faces. The third phase, called structuring, transforms the region into a structure compatible with the structure of the features represented by the feature grammar. The fourth phase, called Identification, identifies the features in these zones. The fifth phase, called modeling, represents the model by features. The feature modeling system software is developed based on this approach.