Uncalibrated Robot Hand-Eye Coordination Based On Optimized ADRC

Robot hand-eye coordination is to plan robot movements making use of the feedback visual information and to complete tasks of robot visual positioning and tracking. As the development of intelligent robot, uncalibrated robot hand-eye coordination becomes a hotspot of this research area. Active Disturbance Rejection Controller (ADRC) is a kind of controllers that not only automatically detect the system model and real-time effects of outside disturbances but also make the compensations which don't separate internal disturbances from external disturbances but directly detect and compensate total of them. Although its function is powerful, it's difficult to adjust its too many parameters. Thus, adopting the Adaptive Genetic Algorithm (AGA) to solve the problem is proposed and applied successfully in the control of various plants in this thesis. Limitations and some corresponding solutions of linear PID controller are analyzed, the principle and characters of ADRC is introduced and some rules of parameter adjusting are summarized. Camera model is presented, robot kinematics is studied and the forward kinematics function and Jacob matrix of MOTOMAN-SV3 is calculated. AGA and its use in the parameter optimizing of the ADRC is researched. A first-order ADRC is used to control the system of Continuous Stirred Tank Reactor (CSTR) and a simplified first-order ADRC is used to control the sinusoidal movement of the plane two-link manipulator in horizontal direction. Simulation results are satisfactory. The control system model of robot uncalibrated hand-eye coordination based on the AGA optimized ADRC is constructed. Two first-order ADRCs of the same construction are used to realize the uncalibrated visual tracking control of six degree-of-freedom robot and the visual positioning of MOTOMAN-SV3 robot. Experimental results prove the feasibility and validity of the algorithm.