Research On Calibration Method Of Generalized Kinematic Errors For Serial Industrial Robots

Industrial robots are critical automatic processing equipment,and high positioning accuracy is helpful to promote the application of industrial robots in high precision machining and assembly.The traditional model-based kinematic calibration method has a limited effect on improving the accuracy of industrial robots.Therefore,a generalized kinematic error calibration method is needed to be proposed to calibrate the geometric errors and non-geometric errors,reducing the positioning errors of industrial robots and improving their positioning accuracy.In order to search for a calibration method that can significantly reduce the positioning error of industrial robots,this paper studies the calibration of generalized kinematic errors for serial industrial robots.Aiming at the problem that the traditional calibration model can only describe the geometric errors but can not describe the non-geometric error,this paper establishes a generalized kinematic error model based on product of exponentials(POE)formula.In the generalized kinematic error model,we consider the influence of geometric errors and non-geometric errors on joint motion and describe the error terms with polynomials whose independent variable is joint position.There are many parameters in the generalized kinematic error model,which requires enough measurement data in the identification data set.In order to improve the measurement efficiency,this paper uses the quasi-random distribution to generate measurement configurations,and plans the measurement path of industrial robots,so that the spherically-mounted reflector(SMR)is visible to the laser tracker in the measurement process.The unknown parameters in the generalized kinematic error model need to be identified according to the measured data.In this paper,the parameter identification of the model is transformed into a nonlinear least square problem.In order to improve the efficiency and robustness of parameter identification,the Levenberg-Marquardt(L-M)method is used to search for the optimal solution of the model parameters.Finally,according to the generalized kinematic error model,the generalized kinematic errors of the robot are compensated by compensating the joint motion command.In order to improve the robustness of the compensation algorithm,the dogleg method is used to search the joint motion command after compensation iteratively.The number of redundant parameters in the generalized kinematic error model based on POE formula is reduced,which reduces the complexity of the model and is conducive to parameter identification.Experiments carried out on the Efort ER50 robot and the Efort ECR5 robot show that the proposed calibration method has high calibration accuracy.