Visual Servo of Uncalibrated Eye-in-hand System with Time-delay Compensation

For the visual servo system, especially with the low-cost camera, there are problems on time delay which is caused by the difference between image processing period and robot control period. In this paper, the time delay in image processing is mainly discussed as the sampling period of the image processing is much longer than that of robot control. A scheme by the robot's pose and estimated image Jacobian is applied to the eye-in-hand visual servo system to compensate the time delay by estimating the image features on the image plane. In this method, the uncertain time delay is not estimated and no assumption is needed for the time delay. As the system discussed is uncalibrated. a update method is proposed to estimate the image Jacobian in each image processing period. Furthermore, a control scheme is presented and the Lyapunov method is employed to prove asymptotic convergence of the image errors. Finally, Both simulation and experimental results are shown to support the approach in this paper.