Three-Dimensional Object Perception for Automated Assembly by the Structured Light Method using Cone Sections

Abstract

Abstract Intelligent manufacturing processes require intelligent sensor integration. For automated assembly this means integration of three-dimensional scene perception. Existing approaches use or to reconstruct the 3-D objects. In this paper the 3-D objects are reconstructed by their planar surfaces. The extraction of these surfaces is done using a with . The object planes are estimated by intersecting the generating cones for the projected disks and the image ellipses. The new method is especially addressed to problems arising in robotics and assembly tasks, where it is necessary to recognize a priori known objects and to estimate their pose.