The performance of the industrial robot is increased constantly with the development of the modern industry technology. Especially, the positioning accuracy problem is increasingly prominent with wide application of the off-line programming robot. For increasing the positioning accuracy, that is, decreasing the error between the actual pose and nominal pose, this paper utilized the statistical analysis method to improve the identification methods of the kinematic error and the non-geometric error, and proposed the compensation strategy for industrial robot according to the classification of robot error sources.This paper researched the selection of the complete kinematic error parameters for the robot endpoint error description, and it set up the kinematic error parameters model. The joint kinematic parameter error models were set up respectively based on Khalil’s D-H homogeneous coordinate matrix modified by Hayati method, the base coordinate and tool coordinate error models were set up based on Stone model. The position error measurement method, and the coordinate transformation method from the robot coordinate system to the measurement coordinate system were researched, meanwhile, the pose error measurement method through three position points was proposed. Based on the forward kinematics and inverse kinematics, the compensation strategy is presented that kinematic parameter error by forward kinematics compensation method and non-geometric error is compensated by inverse kinematics compensation method.In this article, two identification methods of the kinematic parameter error were presented, that are the screw axis error identification method and the kinematic error parameter model identification method. The map from each joint error onto the endpoint error is put forward, and the endpoint errors induced by the single joint were statisitically analyzed, utilized the statisitical result, the coordinate establish order is obtaind for the identification of the actual kinematic parameter based on screw axis method. We researched the method that multi-point fitting the space circle to obtain the axis direction and axis center, and set up each joint coordinate system in sequence according to the axis signature, moreover, solved the D-H model modified by Hayati method to obtain the actual kinematic parameter in reverse.The unified orientation and position error model identifying the kinematic error parameter is proposed, utilized the pose or position error to identify the kinematic error parameter; and the distance error model identifying the kinematic error parameter is proposed. We researched the extended Jacobian matrix of kinematic error utilized analyzing the linearly dependent column vector in this matrix, and deleting the redundant kinematic error parameter, improved the deleting redundant parameter criterion that delete them ordinal from base coordinate system to tool coordinate system, moreover, we proposed the new criterion that delete the high-precision parameter according to statistical analysis of the endpoint error; we obtain 28 identifiable parameters for pose error and 25 identifiable parameters for position error identifying method. In this article, the singular value decomposition is utilized to improve the computing method that solved the extended Jacobian matrix to identify the kinematic error parameter. The estimation method based on the statistical method is presented and it reduced the measurement number, simultaneously, it ensured effective estimation for solving the extended Jacobian matrix to obtain the kinematic error parameter. According to the result of kinematic error parameter, the pose error, the position error and the distance error compensations were respectively presented, the experiment results show that the mehod utilized the statistical analysis for identifying the kinematic parameter error can improve the positioning accuracy.The criterion evaluating the non-geometric error is established according to the second order kinematic parameter error, and the endpoint errors induced by the second order kinematic parameter error and the non-geometric error were computed and compared, from this result choosing the non-geometric error which needed to compensate. In this article, we compensated the compliance errors are induced by gravity and elasto-static, set up a mapping from the compliance errors of joint onto endpoint error, on the other hand, the main attention is paid to analyze each joint compliance error using single axis rotating by laser tracking system using the statistical analysis. To compensate the compliance error, this paper divides this problem into three sequential subtasks: identifying the robot compliance matrix, computing the compliance error of gravity without external loading, compensating compliance errors of non-geometric error on external loading, the experiment results show that this mehod utilized the statistical analysis to select the joint needed to compensate the compliance errors can improve the positioning accuracy and reduce the endpoint error induced by the external loading, meanwhile, this method set up the foundation of research to compensate the compliance error directly using mechanical structure. |