Neural-network-based robust hybrid force/position controller for a constrained robot manipulator with uncertainties

Abstract

Here, an intelligent hybrid position/force controller is designed for a constrained robot manipulator with contact friction between its end-effector and environment in presence of both large parameter and dynamic uncertainties. The controller includes two major parts. The first part, denoted as the main controller, consists of two closed-loops fulfilling motion tracking and force tracking objectives. The second part, called the tuning controller, is an adaptive neural network controller to compensate for the deficiencies of the model-based controller. The stability of the overall system is guaranteed through the Lyapunov and passivity theorems. The performance of the controller is evaluated using numerical simulations as well as experimental implementation. In the experimental analyses, the proposed controller is implemented on a two-link robot manipulator that interacts with a vertical surface. Results show a significant decrease in tracking error in the presence of uncertainties, owing to use of neural network sub-block.