| The traditional image-based visual servoing(IBVS)technology,generally takes a set of image point features as the visual features,which can be selected from Cartesian coordinate space and polar coordinate space respectively,and then the image features move from the initial position to the desired position by designing the feedback proportional controller.Although the traditional method is robust to system uncertainty and camera calibration errors,there are still some shortcomings:(1)due to the coupling of spatial velocity,the traditional method is serious for the redundant motion of feature trajectories in visual servoing task with rotating motion;(2)the depth in image Jacobian matrix is uncertain;(3)the singularity of image Jacobian matrix;(4)the image feature points may leave the field of view in visual servoing task;(5)The traditional polar-based IB VS method has poor effect on visual servoing task with translational motion and long convergence time.In this paper,the 6-DOF manipulator with monocular eye-in-hand is considered as the research object,aiming at the features of image points in Cartesian coordinate space and polar coordinate space,the corresponding visual servoing control algorithm is studied to improve the defects of the traditional image-based visual servoing control method.The following control methods are proposed:Firstly,a proportional differential-fuzzy sliding mode variable structure(PD-FSMC)hybrid controller for 6-DOF manipulator is proposed to solve the problem of image feature trajectory redundancy in traditional image-based visual servoing control method.The controller uses the advantages of PD method and SMC method to improve the robustness of visual servoing system to feature errors.In order to solve the chattering problem of the SMC method,the fuzzy control(FL)method is introduced,where output gain can make a reasonable feedback according to the real-time movement of the image feature points.Compared with the traditional IB VS method,the convergence speed of visual servoing system is improved.Then,a new image-based online robust model predictive control(RMPC)method of visual servoing is presented,which selects the features of image points in polar coordinate space as visual features.A good and single mapped convex polyhedron structure of visual servoing system is obtained by tensor product(TP)model transformation,and robot physical limitations and visibility constraints of the system are taken as input and output constraints,performing predictive control in infinite time domain,the visual servoing task is realized by solving the convex optimization problem of linear matrix inequalities(LMIs)online,and the feasible solution of LMIs guarantees the asymptotic stability of the system.Since the inverse of image Jacobian matrix is avoided directly,there is no singularity of image Jacobian matrix.Compared with the traditional polar-based IB VS method,the new one not only improves the image feature trajectory,but also increases the convergence speed of the visual servoing system.Finally,in order to verify the feasibility of the visual servoing controller based on online robust model predictive control method,relevant visual servoing experiments are carried out by using an experimental platform built by Denso manipulator and Logitech C310 webcam,and the feasibility and effectiveness of the proposed controller are proved by experiments. |
