Study On The Compliance Control System Of Manipulators Based On Fuzzy Neural Network Impedance Control Method

This dissertation studied the manipulator compliance dynamic model, force analyzing, trajectory tracking, uncertainties compensating, control method of the manipulator compliance system.With the development of robots, space technology and industry technology, Industry robots have been required offering more service. Such as in the process of the robot manufacture, on one hand, the end effector of a robotic arm is required to keep contacting on the contour of the constraint surface in some tasks such as deburring, greeding and so on. On the other hand, the contact force between the end effector of robotic arm and the environment should be controlled. If the force is too strong, the workpiece will be destroyed, on the contrary, if the force is too weak, the goal of the production will not be obtained. So, in this case, it is difficulty to control the manipulator and the manipulator should have compliancy. Especially when the environment is unknown which include two sides: the surface of the environment is unknown or not be described by using mathematics and the material of the environment is unknown, so the stiffness of the environment is unknown. In those tasks, the manipulator should be compliance enough to satisfy the need of industry producing.In view of the above control problem, this dissertation does research step by step. The manipulators compliance motion system is a typical nonlinear robotic system. Thus, in this Dissertation, the dynamic model of the system was established by virtue of the Lagrange's principle. The contact force should be controlled actively in the operating space.Due to the unknown constrain, the contact force will be divided into normalvector and tangent vector. Through analyzing, the normal vector of the contact force is very important while the tangent force can be neglected. According to the signal of the force sensor, we design an on-line adaptive law according to the force tracking error which offer support to the next position control. So, the problem of getting the reference trajectory has been solved.There are model uncertainty, parameter uncertainty and environment uncertainty and so on in the manipulator compliance motion control system. As we known, Neural Network has ability to approach arbitrary nonlinear continuous function, Due to those nonlinear uncertainties, A FBF (fuzzy basis function) neural network has been designed to compensate them in order to improve precision and robust of the control system.Due to the unknown constraint manipulator compliance motion system, not only consider the unknown constraint hyper-surface, but also take stiffness unknown into consideration. Above all, because the model of unknown constraint manipulator compliance motion system is very complex and not easy to be simulated by virtue of the precise model. In this dissertation, the simulation of stiffness unknown has been carried out to verify the effectiveness of the fuzzy neural network impedance force control algorithm. The simulation of regulating properly of FBF neural network verify the effectiveness of FBF neural network.