Adaptive NN based Visual Servoing Control for Robot Manipulator with Field of View Constraints and Dynamic Uncertainties

Abstract

This paper studies the constraint control of a visual servoing system which composed of a 6DOF manipulator with a camera sensor mounted on the end-effector subject to the limitations of the field of view (FOV). In visual servoing control, losing information about the visual features can easily lead to task failure. In this work, a barrier Lyapunov function (BLF) is utilized to design the controller and render certain features continuously stay within the field of view, also guarantee the uniform ultimate boundedness of the closed-loop system. No path planning nor information about the actual system dynamics is required. The Moore-penrose inverse term is introduced to ensure the asymptotic stability of the system. In the presence of system uncertainties, an adaptive neural network is adopted to compensate for uncertainties. The theoretical findings are demonstrated via comparative simulation studies based on a 6 DOF robot.