Douglas Edward Richard Clark

Viewer-centered geometric feature recognition

Computer aided design (CAD) and computer aided manufacturing (CAM) systems are now indispensable tools for all stages of product development. The flexibility and ease of use of these systems has dramatically increased productivity and quality of product while reducing lead times. These advances have been largely achieved by automating individual tasks. At present, these islands of automation are poorly linked. One reason for this is that current computer systems are unable to extract geometric and topological information automatically from solid models that is relevant to the down-stream application. In other words, feature information.The objective of the research reported in this paper was to develop a more generic methodology than heretofore, in order to find the generic protrusion and depression features of a CAD model. The approach taken is one relying on a more human type of analysis, one that is “viewer-centered” as opposed to the object-centered approach of most previous research in this area. The viewer-centered approach to feature recognition described is based on a novel geometric probing or tomographic methodology. A five-step algorithm is described and then applied to a number of components by way of illustration.

Collision detection methodologies for rigid body assembly in a virtual environment

In this paper, a simulated three-dimensional virtual environment is created with a virtual 3D track ball for virtual object control. We propose a new technique called HV Partition to detect accurate collision in the assembly of two polyhedral solids in virtual simulation. This is a solid interference detection methodology achieved by automatically partitioning the object into smaller solid boxes. An important advantage of this methodology compared with other approaches is that it can deal with non-convex objects. This means that mechanical components, represented by non-convex polyhedra, traversing any degree of freedom, can be used in this virtual environment. Using this HV Partition method, the precise interference between two polyhedral solid objects can be found. The HV Partition methodology is applied following initial approximate collision detection using traditional bounding box and bounding sphere methods. The smaller the number of smaller boxes, the quicker is the performance of the collision algorithm. An optimal partition method is also given to reduce the number of smaller boxes in an object.

Mass-Spring Simulation of Deformation in Elastic Sheet Structures

Traditionally, continuous deformable models are represented using control points that are arranged in a grid format. The interaction properties between these points are controlled via a series of interconnections that are used in physically based modeling to represent environmental effects within a simulation. These effects vary from the transfer of forces from one control point (or node) to its neighboring nodes, through to representing real-world forces such as friction and energy damping that affect any object placed in an uncontrolled environment. This paper presents a real-time, computationally inexpensive environment for accurate simulations of sheet materials on a personal computer. The approach described differs from other techniques through its novel use of multilayer sheet structures. The ultimate aim is to incorporate into the environment the capacity to simulate a range of temperatures. A pseudo-immersive window on world (WoW) environment is used to handle the implementation of the real-time, aesthetically accurate deformation algorithm (MaSSE-Mass-Spring Simulation Engine). The motion of the sheet is controlled by simulated gravity and through its interaction with a mouse-pointing device. In addition, the mouse may be used to manipulate the sheet. An obvious application of the environment is centered on mechanical engineering-based real-time simulations of heat-sensitive sheet materials. This would allow for a wide range of applications in virtual manufacturing.

A pseudo-immersive virtual environment—a framework for modelling sheet deformation

This paper presents a real-time, computationally inexpensive environment for accurate simulations of sheet materials on a personal computer. The approach described differs from other techniques through its novel use of multilayer sheet structures. The ultimate aim is to incorporate into the environment the capacity of simulating a range of temperatures. A pseudo-immersive “Window on World” (WoW) environment is used to handle the implementation of the real-time, aesthetically accurate deformation algorithm (MaSSE-Mass-Spring Simulation Engine). The motion of the sheet is controlled by simulated gravity and through its interaction with objects that have been inserted into a virtual room. In addition, the WoW interface is used to adjust environmental parameters dynamically and adjust the scene viewing perspective. An obvious use of the environment is centred on mechanical engineering-based real-time simulations of heat-sensitive sheet materials. This would allow for a wide range of applications in virtual manufacturing including the clothing industry and hostile environments.

Four colouring the vertices of the triangulation of a polygon containing a hole

A simple linear-time algorithm is presented for four-colouring the vertices of a triangulation of a polygon containing a single hole. The algorithm consists of reducing a triangulation by the removal of both polygon and hole ear vertices, if any, followed by the removal of colour-isolated vertices until a 3-coloured tessellation remains. The triangulation is then re-built, using at most four colours. The paper concludes by recognising the similarity between the colouring of triangulations of polygons containing a hole and the colouring of bipartite and permutation graphs.

MEASURING GEOMETRIC COMPLEXITY OF 3D MODELS FOR FEATURE RECOGNITION

Geometric Feature Recognition in Manufacturing is a large field of research in which many different methods and solutions have been proposed, and yet no single agreed definition of a feature exists. Moreover, no measure of how complex a manufacturable component is, in terms of features, has been developed. This paper presents a Feature Complexity Index, the aim of which is to encapsulate generic feature concepts without explicit feature definitions and thereby quantify a concept which human intuition understands effortlessly. One possible application of the index is to compare the efficacy of the wide range of disparate feature recognition systems currently being developed.