Geometric reasoning for recognition of cavities
Abstract
This thesis addresses two important problems: (1) how can the cavities of a mechanical part be correctly recognized from the part's boundary representation? and (2) how can we use the derived knowledge of part cavities in an immersive feature-based computer-aided design (CAD) system to provide meaningful navigation to the user, and manufacturing information to interfacing process planners? Automatic recognition of part cavities is a complex problem because the cavities of a part can interact in complex ways; most of the existing systems are provably incorrect. For the part domain consisting of parts with iso-oriented (no inclinations) cuboidal and cylindrical cavities, we present a new volume decomposition approach to recognize the part cavities. We prove that the cavity interpretations generated by our system are correct. We also present algorithms for incremental cavity recognition; these algorithms reason about the existing part cavities to determine how they are modified by attachment of a new design feature. This reasoning enables us to avoid reinterpretation of the part from scratch after every design operation. We also detail the design of a VR-based CAD system that integrates the part cavity recognition techniques. This system not only presents to the user an intuitive immersive interface for the specification of three-dimensional features, but also directly supports CAD-CAM integration. The system derives and suggests hints regarding the presence of machining features (such as steps, holes, slots, etc.) to an interfacing process planner, by reasoning about the part cavities. Tracking the adjacencies between the part cavities allows the system to also suggest a possible sequence for machining the proposed features. The immersive interface enforces that the part cavities remain accessible to the user in the virtual environment for sketching new features directly on part surfaces. This also guarantees that the part cavities are at least theoretically accessible to machining tools for manufacturing the real part.
Degree
Ph.D.
Advisors
Kashyap, Purdue University.
Subject Area
Electrical engineering|Industrial engineering|Mechanical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.