A Variational Bayesian-Based Simultaneous Localization and Mapping Method for Autonomous Underwater Vehicle Navigation

Simultaneous Localization and Mapping (SLAM) is a well-known solution for mapping realizing autonomous navigation of an Autonomous Underwater Vehicle (AUV) in unknown underwater environments. However, the inaccurate time-varying observation noise will cause filtering divergence reduce accuracy localization feature estimation. In this paper, VB-AUFastSLAM based on unscented-FastSLAM (UFastSLAM) Variational Bayesian (VB) proposed. The UFastSLAM combines unscented particle filter (UPF) Kalman (UKF) to estimate AUV poses features. addition, resist noise, method learning introduced into SLAM framework. Firstly, VB used joint posterior probability path noise. Inverse-Gamma distribution model real-time parameters estimation performed improve accuracy. Secondly, reused update stage enhance performance proposed algorithms are first validated open-source simulation environment. Then, system Inertial Navigation System (INS), Doppler Velocity Log (DVL), single-beam Sonar also built verify effectiveness marine methods can reach 0.742% 0.776% range, respectively, which much better than 1.825% 1.397% traditional methods.