Construction of visual servo controller by Kalman filter and created window

This paper describes a new strategy to realize visual servoing by window algorithm and Kalman filter. In visual servoing, disturbances in the image frame have a significant effect when the desired motion commands of the visual feedback controller is synthesized. To address the above issue, the specified windows which show each feature point in the image frame are introduced. Then the commands are calculated by only the information of the created windows. As a result, the motion to track the target objects are not affected even if the unknown objects appear in the image frame. The performance of the proposed algorithm is confirmed by several experimental results.