Computer vision based robotic system for collecting small industrial products

The goal of this thesis is the development of an efficient, robust and inexpensive system for collection and processing of metal products on various processing stations. The main advantage of the system over dedicated automation machines is its versatility, as it can identify and process many different types of products. The complete system consists of a camera, a robot and a computer with our software. Metal products are placed on a flat surface. The camera takes a photo of the surface, which is transfered to the PC and then processed using computer vision. Our software also transforms the points from the coordinate system of the camera to the coordinate system of the robot. The robot is then controlled using these coordinates. The robot picks up a product from the surface and carries it to the processing station where it is processed further – hole drilling, thread rolling or thread cutting, grinding, etc. After the cycle is finished, the product is put in a container. The cycle is then repeated with the next product