Research On Dynamic Parameter Identification And Feedforward Control Of Articulated Robots

Industrial robot integrates modern manufacturing technology,new material technology and information control technology,which is widely used in industrial production processes,such as handling,welding,assembly,processing,painting and so on.Robotic technology is now moving towards high-speed,high-precision,high-intelligence direction.With the continuous expansion of industrial demand and upgrade,more stringent requirements are attached to robotic control accuracy in modern tasks liking laser welding and laser cutting.In the traditional robot control strategies,since the internal PID regulation of the servo driver ignores all nonlinearities in the motion control of robot,the control strategy that only relies on error feedback control can not guarantee the control accuracy of robot at high speed.However,the feedforward compensation control based on robotic dynamics model can speed up the internal error convergence speed of the servo drive and improve the dynamic response characteristics of robot.Several foreign advanced machine manufacturers have introduced model-based torque compensation technology into their respective robot controllers while in our country a lot of time and effort has been invested to carry out the research on robot dynamics now.Although the theoretical research is more in-depth,most of the research are limited on simulation.Aimed at the typical articulated robots in industrial field and based on existing research results,the dynamic parameter identification scheme is designed and perfected in this paper.And then considering the existing experimental conditions,the scheme designed in this paper is verified experimentally.In addition,the feasibility of the feedforward control scheme based on dynamic model is also verified using the universal simulation platform Simulink.In order to research on dynamic parameter identification and feedforward control of articulated robots,first of all,using Eston's ER16 as the object of study in this paper,the advantages and disadvantages of the traditional methods in constructing robot dynamics model are analyzed.Finally,the iterative Newton-Euler method is selected to construct the dynamic model of ER16.What's more,based on the linearized mathematical theory,the minimum set of parameters is calculated with the help of symPybotics toolkit.In addition,the dynamic model above is validated by Matlab Robotics Toolbox.Then,based on the design goal of the excitated trajectory,a trajectory model based on finite Fourier series is proposed in this paper.And under the constraints of the joint state of the robot ER16,the condition number of the regression matrix in the model is minimized as an optimization target.The optimal trajectory is obtained by multiple screening.Finally,the dynamic parameters are identified by the experiments on the front three axes of ER16.In order to deal with the relevant experimental data,a combination of Butterworth filtering and zero phase digital filtering is designed in this paper,supplemented by RLOESS algorithm.This solution not only effectively filters out the high-frequency noise in the data,but also ensures that the phase is not distorted.Finally,the weighted least squares method is used to identify the dynamic parameters and the space rosette trajectory is designed for parameter verification.Based on the parameters identification scheme,the model-based feedforward control research is carried out in the paper.According to the principle of feedforward control and the time constraints in actual control system,feedforward control simulation verification is done for three-axis system of the articulated robots by means of Simulink platform.In this paper,the comparison simulation experiments with or without feedforward control are carried out and it is proved that the feedforward control scheme based on dynamic model is helpful to improve the trajectory accuracy and the dynamic response characteristics of the articulated robots,also for the future research on feedforward control experiments.