Research On Dynamics Uncertainties And Adaptive Control Methods Of Uncalibrated Visual Servoing Systems

Since the external disturbance, target-operator characteristic changes, hand-eye uncalibration or inaccurate calibration, mechanical wear and some other factors, there are many uncertainties existing in the robot visual servoing system. These uncertainties would result in poor controlling performance of the system, or even cannot meet the control requirements. Based on the IBVS scheme, this paper designed appropriate adaptive controllers with the robot dynamics model, which achieved the point to point control in the robot visual servoing system with uncertainties. The research of this paper can be summarized as follows.(1) For the visual servoing system of the camera uncalibrated and robot joint friction unknown, by selecting appropriate linearizable camera parameters and friction parameters, uncertainties caused by the camera uncalibration and unknown joint friction can be converted to uncertainties of the selected linear parameters. Then the paper designed adaptive parameters updating laws to estimate the camera and friction parameters online real-time and used the obtained estimations for controller design. In particular, the paper used the image error as the output feedback and the joint speed as the state feedback to design the robot joint torque controller based on the robot dynamics model.(2) For the visual servoing system of the camera uncalibrated and robot dynamics model uncertainty, by selecting appropriate robot dynamics parameters that can be linearized, uncertainties caused by the nonlinear robot dynamics model can be converted to uncertainties of the selected linear parameters. Then the paper designed adaptive parameters updating laws to estimate the camera and dynamics parameters online real-time, and used the estimations, the visual signal and the robot joint information for the joint torque controller design that can drive robot directly to complete the point-to-point task.(3) For visual servoing system of the camera uncalibrated and robot dynamics model uncertainty, the design of the robot visual servoing system that based on single feature point used in (2), is extended to system design that based on multi-feature points, which means using multiple feature points to design the robot joint torque controller and adaptive parameters updating laws. In servoing motion, the use of multi-feature points in system design can increase redundant image features and avoid bad effect such as servoing failure caused by feature points blocked or ran out of view field of the camera to some degree. And the use of multi-feature points can also improve the robustness of the control system.(4) In order to verify the feasibility and effectiveness of the control schemes designed in (1),(2), and (3), the paper build the robot kinematics and dynamics model for a3-DOF manipulator and do the MATLAB simulation experiment. The performance shows that the robot, in uncertain environments, can complete control tasks satisfactory, and the schemes have good control effect.