Abstract

Visual servoing has played an important role in automated robotic manufacturing systems. This thesis will focus on this issue and proposes an improved method which includes an ameliorative image pre-processing (IP) algorithm and an amendatory IBVS algorithm As the first contribution, an improved IP algorithm based on the morphological theory has been discussed for the purpose of removing the unexpected speckles and balancing the illumination during the image processing. After this enhancing process, the useful information in the image becomes prominent and can be utilized to extract the accurate image features. Then, an improved IBVS algorithm is therefore further introduced for an eye-in-hand system as the second contribution. This eye-in-hand system includes a 6 Degree of Freedom (DOF) robot and a camera. The improved IBVS algorithm utilizes the image moment as the image features instead of detecting the special points for feature extraction in traditional IBVS. Comparing with traditional IBVS, choosing image moment as the image features can increase the stability of the system and extend the applied range of objects. The obtained image features will then be used to generate the control signals for the robot to track the target object. The Jacobian matrix describing the relationship between the motion of camera and velocity of image features is also discussed, where a new simple method has been proposed for the estimation of depth involved in the Jacobian matrix. In order to decouple the obtained Jacobian matrix for controlling the motion of camera with individual image features, a four stages sequence control has also been introduced to improve the control performance.