An international conference connecting people
in CAD research, education and business
Bookmark and Share
Copyright (C) CAD Solutions, LLC. All rights reserved.
Proceedings of CAD'16, 2016, 327-331
GPU Accelerated Algorithms for CAD / Point Cloud Digitizer Data Registration

Venu Kurella, Allan Spence, McMaster University
Bob Stone, Origin International Inc.

Abstract. Automotive sheet metal stamping supplier production rates approach one part every 15 seconds.  With increasing demand that comprehensive geometric quality conformance information be communicated to the final assembly plant prior to shipment, the associated digitizing of millions of points requires more computationally efficient analysis algorithms. For example, an industrial blue LED snapshot sensor can acquire 1 million points per second. At a 0.1 mm nominal point spacing, for even small part areas, many millions of points need to be registered with the 3D coordinate system of the CAD nominal surfaces. The memory and computing power needed to perform this analysis at part production rates far exceeds the capacity of the conventional personal microcomputers. This paper investigates the alternative of using parallel Graphical Processing Unit (GPU) hardware. Use of this hardware exploits the massively parallel architecture of a GPU to accelerate data intensive computations. Designed for high graphics intensity CAD and gaming, a GPU has a complex memory and processing architecture. Hence effective programming resource allocation and utilization is much more complex.

Keywords. Parallel computing, Graphical Processing Unit, GPU, point cloud registration, point-facet matching

DOI: 10.14733/cadconfP.2016.327-331