Improving Feature Extraction Through Closely-Coupled Integration of Fixture Analysis

Abstract

This paper presents a view of feature extraction as a process that involves consideration of manufacturing tools, processes and the fixtures to be used. This view is implemented in MEDIATOR. Some feature extractors use almost entirely geometric considerations [29], others use process and tool information which is often implicitly encoded in the data structures to help guide the feature extraction process. Our view of feature extraction is similar to the second approach in that process and tool information is also used. However, we take this approach one step further and also use fixture information to determine features. The reason behind this is that a feature is considered relevant because there is a method for producing it in the task domain. The set of possible tools, tool motions and fixtures that can be used is generated during feature extraction in MEDIATOR and selection of specific fixture and details are done later in process planning. Advantages of this method include the ability to use task information to strongly constrain the search for valid features, increased likelihood that features recognized will be manufacturable and directly usable by the process planner, and ease of modifying the feature extractor.