Research On Double-encoder Servo System Oriented To Human-robot Interaction Of Collaborative Robot

With the development of technology and changes in social needs,the forms of various products have been transformed into individualization and quality;the industrial production mode is being transformed into flexible and intelligent manufacturing.Under this background,co-robot,which is flexible and simple to deploy,has gradually become the development detrend of robot.Collision detection and direct teaching function embody the essential characteristics of safety and cooperation of collaborative robots,thus,become the basic functions of realizing natural human-robot interaction.Properties such as low cost and high efficiency brought by the sensorless human-robot interaction have attracted many scholars.In order to promote the research progress of sensorless human-robot interaction,the present thesis takes the double-encoder joints of collaborative robots as the research object and then carries out the study of the double-encoder servo system based on system identification algorithms and harmonic drives modeling technology.An idea that capable of realizing and promoting sensorless collision detection and operation comfort of direct teaching is proposed based on double-encoder servo system.The content of this thesis are as follows:Aiming at the shortcomings of existing model identification algorithms,a novel system identification method based on curve fitting to phase-plane trajectory is proposed.With the help of relay position feedback tests,the proposed method is able to accomplish curve fitting to the phase-plane trajectory utilizing the derived phase-plane trajectory expression and optimization theory.According to the fitted phase-plane trajectory and the value of cost function,the accuracy of the parameter identification results can be quickly and intuitively verified.The proposed method is able to avoid the requirement of acceleration information which is necessary for the least square method and the approximation error which is unavoidable for the description function method.The experiment results show that the proposed method can efficiently achieve accurate and robust parameter identification,which means that it can provide parameters for the design of compensators and disturbance observers,thus,laying the model parameter basis for the double-encoder servo system based collision detection and direct teaching.In order to realize the accurate modeling of harmonic drives,the nonlinear characteristics of harmonic drives have been studied.Since the kinematic error of harmonic drives is periodic with respect to absolute angle displacement,it is modeled as the sum of a series of harmonics firstly.Then a method based on the fast Fourier transform is proposed for analysis and identification of the kinematic error in the "harmonics order field",which makes up for the lack of academic research on the kinematic error.Furthermore,the online compensation experiments of the kinematic error are carried out and the results verify the effectiveness of the proposed analysis method.After the compensation of the kinematic errors,the nonlinear stiffness curve of the harmonic drives is described utilizing cubic polynomial and piecewise linear,which realizes the accurate modeling of harmonic drives.Based on the accurate modeling of harmonic drives,the output torque of harmonic drives can be obtained by the transmission error.Thus,it provides a model basis for the implementation of double-encoder servo system based collision detection and direct teaching.With the help of system identification algorithms and harmonic drives modeling,the study of collision detection and direct teaching based on double-encoder servo system is carried out.Aiming at the problem that the sensitivity of sensorless collision detection is limited by the robot joint friction modeling and dynamic error,the equivalent flexible joint model is proposed based on double-encoder servo system,and the external torque is estimated by the model of the harmonic drives.Thus,sensorless collision detection which is capable of avoiding the influence of joint friction model is realized.For the problem that the flexibility of sensorless direct teaching is affected by the uncertainty of the joint static friction,the signals of double encoders are used to determine the direction of the external force during the starting phase of the direct teaching.Thus,with a force compensation in the corresponding direction,a small external torque can move the robot joint with no need to overcome the maximum static friction.The simulation experiment results show that based on the double-encoder servo system,the sensorless collision detection sensitivity,and the direct teaching operation comfort during the start phase can be improved.