This thesis was created by VBG, a company that manages and owns different industrial companies. One of those industrial companies makes couplings and those couplings need to be painted. To make sure some parts of the couplings are not painted, they contain a mask in the form of a plug. Afterward, this mask needs to be removed.

The goal of this thesis is to prove that this demasking process can be automated. This includes finding the coupling on a rack, grabbing it, and removing the plug. To do this, a solution is created using a UR10e robot and a machine-learning model called YOLOv6. This model was chosen after a thorough literature review of possible models.

The end implementation used the model to first find a coupling. Then the robot would move and the model was again used to find the orientation of the coupling. Then the robot could rotate in the correct position and grab the part. Then the plug could be removed using a rod to push the plug out of the coupling. The machine-learning model was evaluated and the future work of the solution was discussed.