1. Ivan Ćirić, Univerzitet u Nišu, Mašinski fakultet, Serbia
2. Žarko Ćojbašić, Univerzitet u Nišu, Mašinski fakultet, Serbia
3. Vlastimir Nikolic, Univerzitet u Nišu, Mašinski fakultet, Serbia
4. Milica Ćirić, Faculty of Civil Engineering and Architecture in Niš, Serbia
5. Predrag Živković, Univerzitet u Nišu, Mašinski fakultet, Serbia
6. Ivan Pavlović, Univerzitet u Nišu, Mašinski fakultet, Serbia
7. Emina Petrovic, Univerzitet u Nišu, Mašinski fakultet, Serbia
In this paper, the problem of robot vision system is considered as a part of hierarchically highest level of control that is used for object recognition and that allows adequate and reliable robotic grasping. Thus, the application of an intelligent classifier is proposed in the object recognition process for an advanced and robust robot vision system. This novel technique allows 3D object reconstruction with accurate determination of position and orientation of a rigid body that can help in deciding where to grasp object with a robot arm. Classification is performed using features obtained from segmented image after pre-processing, image segmentation and post-processing. Such an image combined with classification results enables accurate 3D object reconstruction that is crucial for grasping problem.
Ključne reči :
Tematska oblast:
Mechatronics and Information Technology
Datum:
02.03.2015.