Research On Kinematics Calibration Method Based On MH80? Robot

China has become the world's largest demand for industrial robots.As the country increases its support for the industrial robot industry,higher requirements are placed on the positioning accuracy of the end of industrial robots.Traditional industrial robots have high repeat positioning accuracy through teaching methods.In order to improve the absolute positioning accuracy of the end of the industrial robot,this paper has carried out the following theoretical research work on the modeling,error parameter identification and error compensation technology in kinematics calibration based on Yaskawa's MH80? robot:First,this paper establishes the MD-H kinematics model of the robot,derives the forward and inverse solutions of the robot kinematics in detail,and verifies the correctness of the kinematics model using Robotics Toolbox simulation.Based on the MD-H model,use the differential method to establish the robot end positioning error model as a preliminary calibration model,design the robot kinematics model error and 100 sets of joint angles,calculate the nominal position and actual position of the robot end according to the forward kinematics solution,and use the least square method to identify kinematics model error,Matlab simulation to compare the robot end positioning error before and after identification,shows the effectiveness and limitations of the least square method.Subsequently,in the process of establishing the positioning error model by the least square method,the high-order differential term needs to be ignored and the kinematic model error cannot be accurately identified.This paper introduces the genetic algorithm to transform the error identification into a multi-parameter optimization problem.Aiming at the characteristics of traditional genetic algorithms that have global optimization capabilities but easily fall into local optimal solutions,this paper proposes genetic tabu search algorithm(GATS)and hybrid genetic simulated annealing algorithm(GASA).The simulation results show that both improved genetic algorithms can be fast jumping out of the local optimal solution,the average positioning error of the robot end before calibration is 9.6875 mm,and the identification result is compensated to the kinematic parameters of the robot controller.GATS and GASA reduce the average positioning error to 0.7412 mm and 0.2607 mm respectively.Finally,in view of the situation that the kinematic parameters of the robot controller cannot be changed after leaving the factory,this paper uses the Newton-Raphson method(N-R)to compensate the joint angle of the robot to improve the positioning accuracy of the robot end.In order to illustrate the effect of the three kinematic error parameter identification methods on the robot end trajectory planning,a circular arc trajectory is designed,the average positioning error before calibration is 5.7560 mm,After the least square method,GATS,GASA were used to identify and compensate respectively,the average positioning error of the robot is reduced to 1.8860 mm,0.9718 mm,0.7641 mm respectively,indicating that the two improved identification algorithms proposed have good identification results.