Application of Variational Bayesian-based Noise Remodeling in AUV Navigation

The precise sensor fusion algorithm is the key part to achieve Autonomous Underwater Vehicle (AUV) navigation, but it’s difficult to estimate the time-varying noise of the sensors on board the vehicle. Existing data fusion algorithms usually set the noise of these sensors as a fixed white Gaussian noise, which is difficult to meet the needs of accurate AUV navigation. In this paper, we model the measurement noise as Inverse-Gamma distribution and Student-t distribution, and use Variational Bayesian method to estimate the noise distribution parameters to resist the time-varying noise and improve the adaptive capability of the algorithms. Two most commonly used nonlinear Kalman filters are improved to verify the effectiveness of the proposed algorithms in AUV navigation. The AUV sea trial data show that the proposed algorithms can effectively improve the navigation capability.