Fast object registration and robotic bin picking

Abstract

Businesses have invested a lot of money into intelligent machine vision, industrial robotics and automation technology. The proposed solution of this paper deals with industrial applications of robotic bin picking. In this paper, a pose estimation approach is introduce to determine the coarse position and rotation of a known object using commonly available image processing tools applied to 3D laser range data. This position and orientation is refined by a combination of the well-known Iterative Closest Points Algorithm with the hierarchical object representation of Progressive Meshes to find objects in a industrial environment. This approach is integrated in an object localization system for industrial robotic bin picking. © 2010 Springer-Verlag.