Industrial robot geometric parameter calibration method based on position vector method

Abstract

The invention discloses an industrial robot geometric parameter calibration method based on a position vector method. The industrial robot geometric parameter calibration method based on the positionvector method comprises the steps that a robot position vector model based on a basis coordinate system oxyz is established, and nominal values of a direction vector and a connection vector in a zerostate of a robot are found from a robot manual; groups of joint angles of the robot are randomly set on a demonstrator according to movement ranges of all joints given by the robot manual, and all thejoints of the robot are controlled to move to the groups of joint angle values qij which have been set; a laser tracker is used for detecting a target which is installed at the tail end G of the robot, and a tail end posture and position data of a measured robot are obtained; and an objective function Ej of the geometric parameter calibration of the robot is established, and a genetic algorithm is used for solving the objective function Ej so as to obtain an optimal solution of the direction vector and the connection vector of the measured robot. According to the industrial robot geometric parameter calibration method based on the position vector method, the problem of singularity caused by parallel and vertical adjacent joints is solved, and meanwhile, an industrial robot geometric parameter calibration error optimization objective function is built based on the model.

Description

technical field [0001] The invention relates to a method for calibrating geometric parameters, in particular to a method for calibrating geometric parameters of industrial robots, and belongs to the fields of precision measurement and computer applications. Background technique [0002] The continuous development of high-end manufacturing has increased the precision requirements for industrial robots, especially in applications such as laser welding, laser cutting, and aerospace. The measurement indicators of industrial robot positioning performance are mainly repeat positioning accuracy and absolute positioning accuracy. At present, the repetitive positioning accuracy of industrial robots can reach 0.02mm to 0.1mm, while the absolute positioning accuracy is only at the millimeter level. The repetitive positioning accuracy of robots is more important in teaching programming. With the beginning of the fourth industrial revolution, teaching programming has gradually been unab...

AI Technical Summary

Problems solved by technology

By adding a rotation parameter, this model makes up for the defect of singularity in the DH model when the rotation axes of adjacent joints of the robot are parallel or close to parallel, but when the two adjacent axes are vertical or close to vertical, the model still has singularity , leading to low accuracy of robot error calibration

In addition, the genetic algorithm used in this invention adopts a floating-point binary method, which requires frequent encoding and decoding. During crossover and mutation operations, the fitness value of each individual in the population needs to be calculated first, and then the crossover and mutation individuals are selected according to the fitness value according to the gambler selection method. , it takes a long time to complete one generation of evolution, the optimization efficiency is low, and the probability of crossover and mutation needs to be determined based on experience, the algorithm has poor robustness

[0004] Based on the above analysis, the main shortcomings of the current research work in related fields are: (1) When the two adjacent axes of the robot are vertical or close to vertical, the existing model has singularity, and the optimization objective function established based on the existing model is relatively It is difficult to meet the continuity and simplicity requirements of robot geometric parameter calibration for the model; (2) Robot geometric parameter calibration is a nonlinear optimization problem with complex constraints. The genetic algorithm not only needs to determine the control parameters such as crossover and mutation probability based on experience, but also needs to frequently encode, decode and calculate the fitness value of the individual in the process of selection, crossover and mutation execution. The algorithm optimization efficiency is low and the robustness is poor.

Images