Based On Image Uncalibrated Mixing Binocular Visual Servo For Manipulator

Robot visual servo is an important part in the field of robot research. Its main core is processing the feedback information from the visual system, to drive the robot movement to complete control tasks, such as positioning and tracking. Robot is a highly nonlinear system. And there are also nonlinear mapping relation inside of the visual system. Therefore, the difficult is lie in establishing the mapping relationship between robot and visual system.This paper based on the method of image Jacobian matrix to describe the mapping relationship between the above two, adopted the layout of "eye to hand" and "eye in hand", combining with points and angles as image features to constitute of image Jacobian matrix. Adopting the method that based on image visual servo, combination with Kalman filter recursive online identification image Jacobian. Using the least squares method to estimate the initial value of image Jacobian matrix, as the basis of online update. The main work of this paper is as follows:(1) This paper proposed a method which based on image uncalibrated mixed binocular of six degree-of-freedom manipulator visual servo control.For the pose of manipulator in the task space positioning, this paper using the control method based on image in visual servo closed loop. Kalman filtering is used to recursive identification online image Jacobian. Adopt the Jacobian provided by Kalman filtering guiding the manipulator realize positioning in the task space. A PID controller was designed for the manipulator. The condition for the stability of the controller are analyzed.(2) Applying Australian scientist Peter Corke's open source robotics toolkit(version9.9), combined with machine vision toolkit(version 3.3), established visual servo in matlab2014 a resources simulation platform. With eye in hand and eye to hand camera configuration layout, uncalibrated manipulator visual servo simulation contrast experiment was carried out.(3) Using Denso VS 6556 GM six degrees of freedom industrial manipulator, combining RC7 M controller and mercury series industrial camera equipment to establish platform. Adopting point and angel as the image characteristics, with a constant image Jacobian combining with angular limit, experiment was conducted in the physical platform.Simulation experiment results show that the proposed method, which based on image uncalibrated mixing binocular visual servo control, can effectively and accurately completed six degrees of freedom manipulator posture positioning tasks. The feasibility and effectiveness of the method is validated. At the same time, the method adopted in the physical experiment has good implementation in manipulator posture positioning task.