Robot end-effector 2D visual positioning using neural networks

Abstract

A visual positioning controller of robot manipulator system with a camera in hand is presented in this paper, where a feedforward neural network is involved to drive the end-effector of manipulator to the desired position instead of the proportional controller. In this case, the visual sensory input is directly translated to world actuator domain. Simulation results show that this method can drive the static positioning error to zero quickly and keep good dynamic response at the same time compared with proportional control law.