A Novel State Estimation Algorithm Based on Variational Bayesians Method Applied to AUV

The MEMS-grade inertial sensors are mainly used to provide navigation information because the sensors loaded in autonomous underwater vehicle (AUV) is very stringent. However, the marine environment inevitably leads to larger errors for MEMS sensors due to its complexity and change. The measurement noise, playing an important role in state estimation with high accuracy, is hard to be obtained. To deal with the problem mentioned above, a new algorithm which fuses variational Bayesians into nonlinear filtering is proposed. The proposed variational Bayesians state estimation (VBSE) algorithm not only can obtain the accuracy measurement noise, but also the dimension of the observation vector is augmented to obtain the position information at the discrete time. The real underwater experiments are performed and experiment results give a draw that the proposed VBSE performs in aspect of positioning accuracy of AUV and robustness than the traditional algorithms.