Visual Servoing of Robot Manipulator Based on Second Order Sliding Observer and Neural Compensation

In this paper, we propose a PD-like visual servoing with second order sliding mode observer and neural network compensation algorithm for planar robot manipulators with only position measurement. This controller is designed based on a combining of a PD controller, neural network compensation and a velocity observer. First, a PD controller is designed as a nominal controller to control robot. Then, in order to compensate the uncertainties, an online learning neural network is designed. Furthermore, the controller incorporates a super-twisting second-order sliding mode observer for estimating the joint velocities; therefore, the velocity measurement is not required. The stability of the closed-loop controller-observer is proved based on the Lyapunov method. Finally, a computer simulation results are presented to evaluate the proposed controller.