Some Researches On Trajectory Tracking Problems Of Robot Manipulators

Robot manipulators have been widely used in various fields such as industrial production,argriculture,mechanical,etc.With the development of modern science and technology,the task of robot manipulators is becoming more and more diversified and complicated,and the operation is not limited to repetitive positioning operation,the need for realtime tracking of a given trajectory is more and more urgent.In order to satisfy increasing complexity and diversity of manipulator tasks,research on trajectory tracking control problem for robot manipulators can realize high precision tracking of a given trajectory in a whole process,which is of great practical significance.Robot system is a high order,high nonlinear,strongly coupled complex system with multi-inputs and multi-outputs,and robot task discription changes with different applications.These problems above put forward higher reqirements for the design of high precision trajectory tracking controller for robot manipulators.This thesis studies high performance tracking strategies for robot manipulators with uncertain kinematics?dynamics and friction nonlinearity,distributed multiple robot systems and coordinative multiple robot systems,respectively,the main research is as follows:(1)To reduce the impact to the manipulator system caused by uncertain dynamic parameters and unknown friction in the joint space tracking problem,a fuzzy-neural-network controller is developed based on a fuzzy-neural-network system.Dynamic LuGre friction is introduced to establish the robot manipulator model.A fuzzy-neural-network system is used to approaximate the nonlinear parts including LuGre friction,in the condition that system parameters,load and the average deflection of the bristles of LuGre friction are unknown.The stability of the closed-loop system is proved by Lyapunov theory.Simulation results show that the proposed controller can not only compensate the influence of nonlinearity including friction,but also be rebust to the change of load.High-precision tracking for robot manipoulator in joint space is achieved.(2)Considering the existence of uncertain kinematics,dynamics and singularities in task-space tracking problem for robot manipulators,a task-space tracking stategy is proposed using a fuzzy-neural-network system.The case of known kinematic papameters and invertible Jacobian matrix is first considered.Then the cases that uncertain kinematic parametes and measurement of end-effector velocity is easily polluted by noise are discussed,kinematic parameters are identified online without pricise measurement of end-effector velocity,a fuzzy-neural-network based controller is further designed.Finally for singularity,a variable coefficient of the damped least squares method(VCDLS)is used to solve the inverse kinematics in the vicinity of singularity,inverse kinematics accuracy is ensured and velocity jump is reduced,a tracking controller is built using VCDLS.The stability of the closed loop system is proved by Lyapunov theory.Simulation results verify the effectiveness of the proposed control schemes.(3)An velocity observer based adaptive fuzzy-neural-network controller(AFNNC)is proposed considering task-space tracking problem for networked multiple robot manipulators under distributed topologies.A velocity observer is built when only a subset of the following manipulators has access to the desired trajectories.By selecting the observer parameters according to the derived sufficient conditions,observation of desired trajectory of each manipulator is realized,and at the same time,continuity of observed velocity and acceleration are guaranteed.Using the observer,distributed AFNNC is constructed.Via Lyapunov and grahp theory,the stability the closed-loop system is proved.Simulation results show that the proposed controller can not only achieve high precision tracking,but also be robust to communication failure.(4)A fuzzy-nerual-network based hybrid position-force controller is developed for coordinative multiple manipulators.The dynamic model of entire system is first built,and the task space is decomposed into two mutually orthogonal subspaces using the idea of hybrid position-force control,controllers are designed separately in both spaces,and a coordinative control item is introduced to reduce the mutual influence between the robot manipulators,The stability of the closed-loop system is analyzed by Lyapunov theory.The simulation results illstrate the effectiveness of the proposed control strategy.(5)A co-simulation platform is established using the ADAMS and MATLAB software.virtual prototypes are first established in ADAMS,and then control systems are built in MATLAB/SIMULINK.The joint-space tracking controller,task-space tracking controller,distributed consensus tracking controller and coordinative hybrid position-force controller are used to control each virtual prototype to test the performance of the proposed controllers.