Research On Kinematics Calibration Method For Industrial Robots

Industrial robots are the key equipment in the field of intelligent manufacturing,and the pose performance accuracy of robots is an important technical indicator that determines its application in industrial scenarios.The structural parameter errors introduced by the manufacturing and assembly process of key components are the main reason for the decrease of the pose accuracy,while the calibration for the robotic structural parameters is an effective technical means to improve its pose accuracy.Previously,there have been a lot of researches on the kinematic calibration for industrial robots.The most commonly used method is to identify the robotic structural parameters by minimizing squares of the absolute pose error at the end-effector.However,there are two problems with this method:(1)For applications in different industrial scenarios,the performance indicators that industrial robots pay attention to are not necessarily absolute accuracy.For example,in applications such as machining and welding,the relative accuracy of the succeeding configurations is more important,while in applications such as dual-arm assembly,more attention should be paid to the collaboration accuracy.(2)Due to the non-homogeneity between position and orientation,inappropriate weight coefficients will cause the identification process to fail to converge.Besides,the description of the robotic orientation depends on the reference coordinate system,and different establishment strategies of the reference coordinate system will also affect the calibration results.Therefore,this paper proposes a robust robot kinematic calibration method for the specific application requirements of the industrial site,considering multiple performance indicators.This method minimizes the corresponding maximum positioning error of the three target points at the end-effector of robots to ensure its both positioning and orientation accuracy.Specifically,the main tasks completed in this article are:Firstly,for the situation where the identification process does not converge with the large initial deviation of the parameters to be identified,an MD-H model for robots with linear and non-linear parameters separated is established.Analyzing the structural parameter characteristics of the MD-H kinematics model,the original problem of structural parameter identification for robots can be converted into a separable nonlinear least squares problem:performing linear least squares on the linear part,and further solving the low-order nonlinear least squares problem.By searching for the optimal structural parameters,the objective function can quickly converge to the global minimum,expanding the range of parameter convergence,improving the convergence speed and enhancing the robustness of the identification process.Besides,the influence of excessive initial deviation of linear parameters can be reduced on the convergence of the algorithm.Secondly,in view of the reduction ratio and coupling ratio errors of the robotic drive train,such as timing belts and gear reducers,a kinematics model at the joint level can be established,which takes into account the transmission ratio errors;Under the heavy-load condition,the torsional deformation of each joint due to the balance of the external torque is considered,and an approximate linear deformation model of the key components of the joint drive system is established.With the acquired pose accuracy data at different configurations,the transmission ratio errors and static stiffness coefficients of each joint axis can be identified,and the robot control law considering the transmission ratio parameters and joint stiffness coefficients is designed and updated to improve the robotic pose accuracy.Thirdly,for the applications of industrial robots in different scenarios such as automobile manufacturing,machining,loading and unloading,grinding and polishing,handling and palletizing,assembly,spraying,etc.,the relative accuracy corresponding to the sequential configurations of a single robotic arm and the collaborative accuracy between multiple robotic arms are defined.This paper also proposes methods to identify the structural parameters of industrial robots in consideration of absolute,relative and collaborative accuracy indicators to improve the performance under specific working conditions.Finally,in view of the issue of the non-homogeneity of the pose representation and the dependence of the reference coordinate,the pose information of the end-effector is obtained by measuring the positions of the three target balls fixed at the end-effector,and the performance indicators corresponding to each target ball are calculated to construct the minimax optimization problem,by optimizing the worst performance indicator corresponding to the three target balls.Through the linear approximation of the nonlinear model,the structural parameters ensuring both the positioning and orientation accuracy could be iteratively identified.What's more,the introduction of the minimax algorithm can increase the robustness of the identification process.