Research On Zero-force Control Of Collaborative Robots Based On Dynamic Parameters Identification

In the context of “Made in China 2025”,human-robot collaboration has gradually become an important issue in robotics research,where humans and robots need to share a workspace and interact safely,and collaborative robots are born.Zero-force control is an important technology of collaborative robots,which allows the operator to drag the robot freely to perform teaching work without the need to program a trainer,significantly improving the efficiency of the robot.However,the capability of zero-force control currently relies heavily on external sensors,and the capability of zero-force control without external sensors is highly depends on the accuracy of the dynamic model.In this paper,collaborative robots was used as research objects,the parameters of the dynamic model of the robot and the payload were identied respectively,and the zero-force control of the robot without external sensors was realized.The details of the study are as follows:First of all,the kinematic model and Jacobian matrix of the collaborative robot was analyzed,and the dynamic model was derived based on the Newton-Euler formulation.The effect of friction was considered in the dynamic model and an improved friction model was developed based on the Coulomb-viscous friction model through single joint friction experiments.In order to facilitate the realization of the subsequent dynamic parameters identification,the linearization of the dynamic model and the minimum parameter set was carried out with the help of sym Pybotics toolbox,and the accuracy of the derived dynamic model was verified by the joint simulation of ADAMS-Simulink.Secondly,in order to better excite the dynamics of the robot during the experiments of dynamic parameters identification,an excitation trajectory based on Fourier series was designed,and the excitation trajectory was optimized by using genetic algorithm.To address the problem of low accuracy of dynamic parameters identification,the traditional identification algorithm was improved,and an iterative reweighted least squares identification algorithm was designed and the parameters identification was completed.The results show that the identification algorithm can accurately identify the dynamic parameters,and compared with the identification results of the traditional identification algorithm,the root mean square of the deviation between the predicted and actual moments of the six joints is reduced by 15.98%,which improves the identification accuracy of the dynamic parameters.Then,considering the effect of payload on the robot joint torque,the dynamic model was derived derived for the payload at the end of the robot,and the accuracy of the payload dynamic model was verified by using a joint simulation method.Analyzed the shortcomings of the traditional payload identification method,a payload identification method without robot body dynamic parameters was designed,and the parameters identification for two kinds of payloads were completed.The deviation of the payload mass and the root mean square of the deviation of the actual torque from the predicted torque were used as evaluation indexes,respectively.The results show that the payload identification method in this paper is fast and accurate,which can provide theoretical support for the subsequent implementation of zero-force control when the robot with payloads.Finally,the impedance control model was analyzed theoretically and the impedance control equations in joint space and Cartesian space were derived.Based on the identification of robot dynamic parameters and payload dynamic parameters,the zero-force control algorithm was designed by combining the characteristics of impedance control model,and the friction model was optimized for the starting problem.The experimental results show that the maximum value of the external force does not exceed 35 N and the trajectory deviation of the Cartesian space does not exceed 4mm during the dragging process,both of which meet the robot teaching requirements and verify the feasibility of the zero-force control algorithm in this paper.The parameters identification method proposed in this paper can effectively improve the identification accuracy of dynamic parameters,and the zero-force control algorithm designed in this paper can realize the direct teaching of collaborative robot without relying on external sensors,which can improve the efficiency of robots and has guiding significance to promote the development of human-robot collaboration technology.