Robust Dissipative Hybrid Position/Force Control Of Robot

Dissipation theory is an ideas which we make use to analyze and design the control system on the view of energy, and has a wide range of applications in the study of stability of the nonlinear system. Robotic systems are very typical nonlinear systems and contain many uncertain factors which have great impact on the performance of the controller. Therefore, addressing this issue two kinds of intelligent control strategies are proposed in the force control loop based on the robot hybrid force/position control in the paper.The first is the passive force control of robot. The robots system are divided into two parts: nominal model and uncertainties, then we design the two controllers separately. For the nominal model, the feedback controller is designed to make it passive, and the effectiveness of the controller is proved by two ways based on Lyapunov theory and KYP (Kalman-Yacubovitch-Popov) lemma. Against the uncertainties of system, we assume a known upper bound of uncertain items, then design a controller according to the bound to offset uncertain items. The simulations prove the validity of the compensation controller.The second is H∞robust force control of robot based on neural network. H∞robust control makes use of the controller whose structure doesn't change to offset the impact of uncertainties, so that the impact of the disturbance of system on the output is inhibited at the required level. Designing controllers according to the relationship between the HJI inequality and L2 gain theorem can avoid directly solving HJI inequality, and simplify the process of designing controllers. As it is necessary to know the upper bound of uncertainties for the H∞robust control, so improved the RAN neural network with features of online learning is used to learn and approximate the upper bound of uncertain items. Then robust controller eliminate the approximation errors, thus the controller getted can guarantee the stability of the system. The two combined can make the system get a good transient performance.