Series-parallel double-arm transfer robot kinematics parameter accurate calibration method based on UPF-RBF combination model

Abstract

The invention provides a series-parallel double-arm transfer robot kinematics parameter accurate calibration method based on a UPF-RBF combined model, and the method comprises the steps: firstly constructing a kinematics model and a kinematics geometric error identification model, and then measuring the real position of an end effector through a three-dimensional laser tracker; the position error vector is input into a UKF algorithm to obtain a geometric parameter deviation initial value, and the UPF takes a UKF parameter identification result as a priori condition and inputs the UKF parameter identification result into the PF for robot geometric parameter re-calibration to preliminarily improve the positioning precision of the hybrid double-arm robot; designing an RBF neural network model for indirect compensation for non-geometric errors with interleaving coupling and strong randomness; a self-adaptive link is designed for online correction according to the situation that variable value disturbance exists in actual carrying operation; compared with a traditional parameter calibration method, the combined model provided by the invention effectively improves the positioning precision of the series-parallel double-arm robot, and has certain self-adaptive adjustment capability.

Description

technical field [0001] The invention belongs to the technical field of industrial multi-arm robot modeling and parameter calibration, in particular to a method for accurately calibrating kinematic parameters of a hybrid double-arm handling robot based on a UPF-RBF combined model. Background technique [0002] The development of industrial automation has put forward higher requirements for the positioning accuracy of industrial robots. At present, the main methods for the calibration of robot kinematic parameters include least squares method, maximum likelihood estimation, genetic algorithm, etc. Among them, the least squares method is simple in operation and fast in convergence, but it is easy to fall into the problem of singularity of the generalized inverse matrix, which has certain limitations. The maximum likelihood estimation has high identification accuracy and is greatly affected by the selection of initial values, which can easily lead to local convergence. Genetic ...

AI Technical Summary

Problems solved by technology

[0005] In addition, due to the heavy weight of the industrial hybrid dual-arm robot, it is easy to cause flexible deformation of the joints and elastic deformation of the connecting rod during heavy load operations, and long-term use will cause mechanical wear and changes in gear backlash, resulting in positioning Reduced accuracy

These non-geometric error sources are complex and diverse, and it is difficult to be accurately and completely modeled

Images