Research On Modular Joint Control System Based On Flexible Modeling And Compensation

Traditional robots are mostly developed for specialized tasks,and have a very limited application scope.When the task is changed,or the surrounding environment changes,it is often required to develop a new robot,which need long development cycle and high cost.In addition,with the development of robot technology and the expansion of the scope of application of the robot,sometimes people can not fully predict the application environment and tasks of the robot.In this context,the researchers started the research of modular robots.Modular robot has been one of the hot topics in the robostics field study with reducing the development cycle and expand the application range of the robot.Modular joint,as the basic structure of modular robot,its control performance will largely determine the control effect of the whole robot.In the control system of the joint,the application of the harmonic reducer and torque sensor,brings the flexibility that can not be ignored,limits the ability of the joint to the high precision and high frequency task.Therefore,in the modeling and control of the joint,it is needed to consider the flexibility in the design of the joint control system.In this paper,the research of modular joint control system based on flexible modeling and compensation is studied as follows:Firstly,the design of the modular joint distributed control system is carried out.A distributed control system based on CAN bus has been adopted as the modular joint robot control scheme.On this basis,the overall framework of a single joint's control system is determined,and the development of the modular joint embedded control system is carried out.The control of the current,velocity and position loop of the joint is implemented.Secondly,the flexible joint is modeled.The influence factors of precision including joint internal flexibility,friction,clearance,disturbance and other factors have been analyzed.Then the analysis and modeling are made from three aspects:the motor,the frictional force and the joint flexibility,and the integral modeling of the joint is completed.In the next place,the parameter identification for the nonlinear friction model of the joint is achieved.According to the established friction model,the model parameter identification experiment has been designed.After obtaining the experimental data,the experimental data were model fitted by the genetic algorithm.After the fitting,compare the model fitting results with the Stribeck model and the disperse LuGre model and the validity of the friction model is verified.Finally,the design of the flexible joint control method is accomplished.On the basis of compensating friction,the control of the flexible joint is controlled by singular perturbation,while in view of the shortcomings of the method,the flexible compensation of the joints is made.On the basis of singular perturbation,employing nominal model based sliding mode control into slow subsystem,and then the effectiveness of the proposed hybrid control rate is verified by means of simulation experiment.