Image Jacobian Matrix Research And Application Based On Adaptive Cubature Filter

Uncalibrated visual servoing is an important research direction in the field of robotics.The main problem in uncalibrated visual servoing is how to obtain the image Jacobian matrix(IJM)that reflects the mapping relationship between the image space and the robot operation space.Based on the published literatures,this thesis mainly studies the estimates of image Jacobian matrix and moving target state.The main works can be described as follows:An IJM online estimator based on adaptive cubature kalman filter is designed.Since the image Jacobian matrix is a nonlinear system and the statistical characteristics of actual noise can not be fully known,an IJM online estimator based on adaptive cubature kalman filter with high approximation precision of nonlinearity is designed.In the IJM online estimator,the state vector is an augmented vector formed by concatenations of the row elements of the IJM,while the observer vector is the noisy image feature.The experimental results show that the research in this thesis improves the accuracy of uncalibrated visual servoing system.Aiming at the moving target state estimation of hand-in-eye configuration,a strong tracking kalman observer is designed.For fixed observation point,the classic strong tracking kalman filter(STKF)observer has a strong ability to track the mutation state.However,under the hand-in-eye configuration,the camera(observation point)is movable.Thus,the observation vector 1s modified to ensure that the STKF observer can be applied to hand-in-eye configuration.The simulation and physical experiment platforms of the uncalibrated visual servoing system are set up.The simulation experiment platform is built on the redevelopment of Peter Corke's Robotics toolbox.The hardware of physical experiment platform is composed by MOTOMAN-SV3X,Basler camera,and Advantech industrial personal computer.The software platform is developed by C++ language in VS2013.Finally,the relevant experiments are carried out on the simulation and the physical platforms.The experimental results show that the proposed algorithm has high accuracy and can be used for improving the control level of the uncalibrated visual servoing system.