Abstract: A nonlinear mapping model between image and robot space for the robotic calibration-free hand-eye coordination problem is analyzed, and it is viewed as system unmodeled dynamics. Based on the principle of Auto Disturbances Rejection Controller for dynamical system with uncertainty, the robotic calibration-free hand-eye coordination controller is designed through compensating for the non-modeling dynamics and disturbance. This kind of controller, unlike other methods, is independent of specific tasks, and provides a unified approach for uncalibrated robotic visual servoing. Simulation and experiment results demonstrate the effectiveness of this method.