Grinding Ball Recognition and Positioning of Automatic Ball-Adding System Based on Visual Robot
During the casting process,numerous pits and protruding structures left by the casting mouth are randomly distributed on the surface of the grinding ball,resulting in many grinding balls that cannot be recognized by the automatic ball-adding system based on visual robots when using matching methods such as surface shape or grayscale correlation to identify and locate them.In addition,there is a height difference on the surface of the grinding ball pile,and the grinding balls at low positions may be identified and located earlier than those at high positions.When the electromagnet is driven by the robot to suck up the grinding balls,it will collide with the grinding balls at high positions and be damaged.In response to the above situation,a layered threshold segmentation method is adopted,ignoring the details of the grinding ball surface.Only the minimum circumcircle of the segmented independent area needs to be identified,and the coordinates of the grinding ball vertex corresponding to its center perpendicular to it need to be calculated to achieve the positioning of the grinding ball.The results show that the problems of the grinding ball not being recognized and the electromagnet being damaged by collision has been effectively solved,ensuring the reliable implementation of the automatic ball-adding system function.
NETL
NSTL
万方数据
