Intelligent picking of chaotically stored objects

A robot system for the intelligent picking of chaotically stored objects by the analysis of depth data is presented in this research article. This "bin-picking" enables the automatic storing of objects which is a prerequisite for the automatic supply of a processing machine. In order to improve conventional localization systems, special attention is paid- apart from the identification of object positions-to gripping point calculation for the picking and to collision avoidance. To adapt the system to different object geometries, an approach that aims at storing the necessary object data in an off-line generated data base was chosen