The task of this master's thesis is to examine the possibilities of application of 3D stereo vision system on an industrial robot in an unstructured working environment. At the beginning, the spatial arrangement of the basic elements of the robotic station was designed, which consists of a Fanuc industrial robot (Fanuc robotic arm), a Fanuc 3D stereo vision system and two tables with boxes containing work items. After that, a standard process of calibration of the vision system and robotic tools was performed, and in the robot’s programming interface the learning of the vision process was performed, which requires the position and spatial orientation of various working objects that have been arbitrarily placed in the box. Based on the learned vision processes, the Fanuc TP program was created that recognizes and localizes various objects in the box, gripping them with a vacuum gripper and performs separate palletizing for each type of object. Also, as part of this master's thesis, the accuracy and repeatability of the 3D stereo vision system was tested and its work in different light conditions was verified. Finally, a simulation of a robotic station with a 3D stereo vision system was created in Fanuc's Roboguide simulation software package. The complete experimental validation of the workpiece handling setup in an unstructured working environment was performed on the Fanuc M-710iC/45M industrial robot in the Autonomous Systems Laboratory at CRTA.