Assessing the potential of computer vision for precise workpiece positioning in drilling applicationsKorošec, Niko (Avtor) Karner, Timi (Mentor) Hace, Aleš (Mentor) Nilsson, Sofie (Komentor) računalniški vidavtomatizacijaoff-line programiranjeThe thesis addresses the theoretical and practical approach to developing an application for drilling and detecting workpieces. It involves using a UR5e collaborative robot, equipped with a drill and a camera that detects the workpiece, determines its position and orientation, and then communicates this information to the robot, which moves to align the centre of the workpiece with the TCP of the camera. The robot then drills into all four edges of the detected workpiece. The development of the application was carried out using RoboDK, a tool for offline programming of robots, and the RoboDK Python API, which handled the entire image processing and program logic. The research primarily focused on the deviations between absolutely fixed and randomly positioned workpieces. The application holds great significance in the industry, as workpieces often do not have a consistent position since they are frequently thrown onto the work surface or conveyor belt. With this application, this inconsistency is not an issue, as the detection of workpieces and machine vision allows us to easily identify the workpiece and perform the required operation.[N. Korošec]20252025-01-27 13:33:45Diplomsko delo/naloga91678sl