Kinematic Parameters Identification Of Robots And Research On The Redundant Parameters

The end of robot needs high accuracy, but the error in the process ofproduction and assembly, the joint clearance and flexibility of arm reduce theaccuracy of robot arm end, in order to improve the accuracy, it is necessary toidentify the kinematic parameters. In this paper, with chain robot as research object,the kinematic parameters and redundancy parameters are identified based onkinematic model and the kinematic error model, the study has theoretical referencevalue and engineering application value of parameters identification.Kinematic model is build based on DH method. The relationship betweenrobot end position and DH parameters is obtained and Jacobian matrix forparameter identification is the coefficient matrix of error model. The transformationrelation of differential variables in different coordinate systems and thetransformation relation of DH parameter error and attitude error of joint coordinatesystem are deduced, further error model of robot end pose is obtained.With6DOF robot as an example, using least square method identifyparameters based on DH model and MDH model, the other kinematic parametersexcept forΔd₅, Δθ₅, Δa₅, Δa₅, Δθ₆, Δα₆have been given a more preciseidentification result, parameters significantly reduced after the end of the positionerror correction, and the identification results of MDH parameter were better thanthat of DH parameters. The redundancy parametric research is discussed andrealized the existence of redundant parameters will seriously affect the accuracyand robustness of parameter identification.Redundancy of generalized error parameters and DH error parameters wereanalyzed respectively. The linear relationship of relevant column of Jacobian matrixis deduced by theoretical method, obtained the generalized error parameters and DHerror redundancy parameters of different robot configurations （different kinematicparameters） at different identification conditions （measured end pose or endposition only）. With the6DOF robot as an example mentioned previously, theresults of redundant parametric research is simulated and verified, andidentification results after removing redundant parameter were better than thatbefore removing redundant parameter, all the kinematic parameters converge on thereal value and the results of redundant parameter study is verified, so this study hassome reference value.