Effect of measurement noise on the performance of a depth and pitch controller using the model predictive control method

Abstract

In this paper a depth and pitch controller for a hover-capable AUV is designed and implemented in simulation. The effect on controller performance of random Gaussian noise on the feedback signals is evaluated. It has been shown that very small levels of measurement noise will result in the controller performance degrading substantially and behaving in an erratic fashion. A polynomial type filter has been proposed and integrated into the model predictive control algorithm. This modification reduces the effect of the measurement noise substantially and improves controller performance.