Kalman filter visual servoing control law

Abstract

This paper introduces a Kalman filter based visual servoing control method that reduces noise sensitivity. It is shown to be completely controllable and observable under certain mild conditions. Visual servoing simulations are performed for a six-axis robot manipulator with both moving and static targets. The controller, if tuned properly, yields equivalent performance to Gauss-Newton for low-noise scenarios and improved performance in the presence of increased camera noise.