| With the advantages of fast operation,high efficiency,modular structure design,flexible control system and high repeatability,industrial robots play an increasingly important part in advanced automatic production and intelligent manufacturing.In particular,the robots are playing an irreplaceable role in high-precision,high-strength,high-risk projects.With the continuous expansion of the application field of industrial robots,higher and higher requirements are put forward for the working accuracy and performance of the robot.There are many factors that affect the working accuracy of the robots,and the error comes not only from the robot parts processing assembly and motion control algorithm,but also from the collision,wear,elastic or inelastic deformation during the use of the robot.The robot calibration and error compensation are effective means to improve the working accuracy of the robot.Based on the application requirements of off-line programming of high-precision tasks,the robot multi-mode calibration technique and the multi-source error modeling method are studied through the combination of simulation and experiment.The prediction and compensation methods of trajectory error and nonlinear error are also investigated in this work,so as to improve the method of improving robot accuracy and optimize off-line programming operations to the maximum extent.(1)The inverse kinematics problem of robot with geometric parameter errors is analyzed,an optimal selection algorithm of complete analytic solutions and a joint small increment method based on the optimal analytical solution and improved Jacobi iteration are proposed respectively,which can be used as an efficient calculation method for real-time control of robot.Based on the principle of differential motion and the theory of partial differential equation,the position error and orientation error of the robot under the influence of geometric parameter errors are analyzed.For the non-geometric error,the sensitivity of the robot positioning error to the change of external payload and temperature is discussed by single factor experiment.Then the compliance error model is constructed based on the linear torsion spring theory and the rigid-flexible coupling pose error model of the robot is established.Matlab simulation experiment is used to verify that the modeling accuracy of the 39-parameter error model for the geometric parameter errors and linear deformation errors of the robot reaches 99.9%.Meanwhile,four groups of comparative experiments are designed to analyze the coupling influence between the robot geometric and compliance parameter errors,and the importance of non-kinematic calibration technology for the correct identification of kinematic parameters is proved.(2)The full pose measurement optimization method is studied emphatically.The interference detection problem in the process of measurement is decoupled into system structure interference detection and optical path angle limit detection.The capsules and four-point method are used to simplify the physical structure of the robot and the measuring tool respectively,and then the collision detection algorithm based on the minimum distance method and the angle limit detection algorithm are proposed.Based on the analysis of the identification performance of the measurement pose,combining the observability index O1 and the identification accuracy index ??,an intelligent selection strategy of measurement pose based on IAPSO algorithm is proposed by taking the singularity of the trajectory path as the constraint to remove the inferior particles.The goal of this strategy is to intelligently provide a set of optimal measurement pose configuration with the best number of poses for the measurement process.The Matlab simulation technology is applied to simulate the experimental environment,and the rigid-flexible coupling model of the robot is identified under the measurement disturbance of different sizes.The results show that the intelligent measurement scheme can not only enhance the automation of the robot measurement process,reduce the influence of human error and environmental noise,but also has higher stability and accuracy in parameter identification.(3)According to the object of parameter calibration and the object of precision improvement,a multi-mode calibration method is put forward.To solve the problem of the instability of calibration result,the K fold cross verification method based on generalized performance evaluation is used to obtain the best calibration model,so that the user can flexibly select the desired calibration mode and realize stable and reliable parameter calibration.The extra error caused by the change of the input joint value in the robot controller and the singularity problem of Jacobian matrix are analyzed,and the joint compensation value is obtained stably by using ADLS method and GN method.The perturbation compensation method of Cartesian space pose/position error is proposed by using the minimum residual error as the optimization objective.The continuous change trajectory error of continuous motion trajectory along the motion direction is studied,and the discretization points with time-stamped is used to represent the trajectory.The multi-objective optimization problem in trajectory correction is analyzed,and a piecewise compensation method of trajectory based on Pareto optimal principle and weighted sum algorithm is proposed to realize the comprehensive optimization of the length of sub-trajectory,the angle between adjacent tangents and the compensation rate of prediction error.(4)The correlation between the positioning error and the spatial position of the robot is analyzed by using the experimental analysis methods such as Monte Carlo and control variable method,and the results show that the consistency of orientation is the necessary condition for the feasibility of spatial interpolation compensation.The influence rule of the change of orientation on the position errors of fixed-point is analyzed,and then a fixed-point compensation method based on sinusoidal phase offset error model and vector angle weighted average algorithm is proposed to eliminate the influence of the change of orientation on positioning errors.Thereby a spatial IDSW interpolation compensation method based on uniform data field is established.The nonlinear error existed after calibration or the robot itself is studied,the method of reducing the dimension of DNN model is discussed,and the compliance error caused by the change of external load and the nonlinear deformation of joint caused by large-scale change of torque are analyzed.Then the compliance error model based on the linear piecewise method is established,which can be used to reduce the space and time complexity of the DNN model.With the general table of measurement configuration of orientation,the nonlinear prediction compensation method based on GA-DNN is proposed by using the optimized DNN architecture and training parameters of GA.Thus,the global nonlinear error intelligent prediction compensation method of robot is proposed to meet the requirements of stable and reliable high positioning accuracy of the robot within the range of rated load and workspace.(5)Based on the Matlab development platform,the robot calibration and off-line optimization system is designed and completed,which provides a good software foundation for practical application and subsequent experimental research.The robot calibration and compensation experimental platform is established.Systematic experimental verification is carried out on the robot multi-mode calibration technology,pose error compensation method,trajectory correction mechanism and nonlinear error intelligent prediction compensation method proposed in this paper.The results of comprehensive improvement of accuracy show that for the Staubli RX160 L robot that has not been calibrated by the factory,using optimization kinematic calibration and the global nonlinear error predicted compensation,the progressive accuracy improvement method can improve the absolute positioning accuracy of the robot by about 96.2% to 97.8% compared with the original error.Finally,according to the international standard ISO 9283,the performance of the Staubli RX160 L robot after calibration and compensation is evaluated,including three performance indices that have significant influence on the performance of the robot: pose accuracy and repeatability,distance accuracy and repeatability,trajectory accuracy and repeatability. |
PDF Full Text Request