Optimization and Implementation of the localization and control algorithms in robot arm with visual feedback

Abstract

In this paper, chess-type plane is used as calibration patterns, and the camera is calibrated by utilizing linear camera model with sub-pixel precision, we propose an improved algorithm used to locate the position with the visual information. The new method is based on improved SUSAN detecting and is combined with quick image matching which is used to get the coordinates of the end executor of the manipulation arm in image coordinate system. Based on algorithms proposed, the coordinates of the end executor in the world coordinate system is resolved through the linear camera model which is calibrated. The method improved is used as the visual position to feedback control a 2-dof robot arm system, and give out the actual control compared experimental results.