A knowledge-navigation system for dimensional metrology

Journal of Research of the National Institute of Standards and Technology, March-April, 2002 by Howard T. Moncarz

4. Techniques

The approach and scenario can be implemented with the techniques described below.

4.1 Virtual Objects

The main interface to the knowledge base is a virtual object (in 2D or 3D) which is the representation of a concept. The object is decomposable into sub-objects, reflecting the concept's decomposition, and the sub-objects are linked to information in the knowledge base. However, depending on the context of the concept, the sub-objects will be linked to different information. For example, assuming the concept is the part, the question could be asked, "what is the set of DMIS instructions to inspect one of the part's features?" In this case, the part's context is DMIS, and if a part's feature is selected, a set of DMIS instructions to inspect that feature will be retrieved from the knowledge base and displayed. In other words, for each sub-concept, the corresponding sub-object will be linked to a number of difference pieces of information in the knowledge base, and the information retrieved when that sub-object is selected will be dependent on what context (or type) has been set for the parent object.

The virtual object might represent a physical object, e.g., a machined part; or it might represent an abstract concept, e.g., an inspection process where the separate applications of the process are represented by a set of boxes. For the latter case, the sub-objects are the boxes. However, a machined part is comprised of features that, generally, are created by volume-removal operations. In this case the features are the sub-objects. They may be the volumes removed, e.g., holes, or they may be the converse--what is left after material is removed, e.g., pegs.

The Virtual Reality Modeling Language (VRML) (9) was used to create the virtual objects in the demonstration prototype. VRML is not designed to represent a boolean subtraction (such as a volume removal). Therefore, for a physical object a technique was developed to superimpose a set of transparent surfaces that represent each feature on top of a solid model of the object. The information needed for a particular object is derived from a STEP AP224 file (10), which specifies the individual features of an object as well as the entire object.

When a sub-object is selected, two signals are sent out from the parent object. The first signal identifies the parent object and the second signal identifies the sub-object selected. A second object can be "wired" to receive those signals. That object can take two actions. It can set its information state, e.g., to specify what type of information to display when its own sub-objects are selected; or, it can take an action based on the signal, e.g., to display information based on the signal received and its previously set mode. Finally, all objects have a mode that is set to "MOVE" or "INFO." In the MOVE mode, the object can be manipulated by the mouse (as described below under object manipulation). In the INFO mode, a mouse click on a sub-object will link to information that can be displayed in another window.