Study on uncalibrated algorithm for dynamic visual servoing system

In order to calibrate cameras and robot models precisely in industrial practice, a novel dynamic uncalibrated eye-in-hand visual servoing system to track a moving target is proposed. The method does not require calibration of camera and robot kinematic models. Visual serving algorithm to track dynamic image is developed through minimizing nonlinear objective function. Because the large residual is not approximated in dynamic environment and the change of composite image Jacobian with time increment can´t be computed in visual servoing system now, large residuals are dynamic approximated and the change of composite image Jacobian at each iterative step is computed. Simulation results demonstrate the validity of these approaches.