Fine localization of complex components for bin picking

Abstract

The aim of this paper is to present verified parameters of a particular approach to one sub-procedure of the bin picking problem. To successfully implement an automatic bin picking application using a robotic arm, it is necessary, among others, to detect the precise position and rotation angle of a selected object. In this approach, the procedure is considered as a two step operation. The first step provides an initial guess of both position and rotation angle, while the second one should specify the pose as exactly as required for following operations. The goal of the paper is to determine a correct relation between those two mentioned steps, i.e. to specify the maximal possible degree of uncertainty provided by the first step so that the second step works correctly. The proposed problem is dealt with as an industrial contract and the results are clearly dependent on specific conditions. However, it can be plainly used as a first insight into the problematics of bin picking, and provide a good starting point for deeper investigations.