Bayesian-Inversion-Based Multisensor Fusion Localization Algorithm for AUV
Wanlong Zhao, Kaiyuan Liu, Ruitong Liu, Gongliang Liu, Weixiao Meng · IEEE Internet of Things Journal · 2025
Underwater positioning is a crucial technology for Autonomous Underwater Vehicles (AUVs) to effectively carry out a range of underwater tasks. With an underwater environment that is complex and constantly changing, it can be challenging for a single positioning technology to keep up with long-range, multi-target, and high-precision applications. Solutions in the field of underwater positioning are continually evolving and improving. A multi-sensor fusion positioning scheme, which integrates multiple sensors including the Inertial Navigation System (INS), Long Baseline (LBL), Doppler Velocity Log (DVL), Conductivity Temperature Depth (CTD) and Depth Gauge (DG), can provide more accurate and reliable positioning. In this paper, a Bayesian inversion based Multi-sensor Fusion Localization (BIMFL) algorithm is proposed to enhance the performance of underwater fusion positioning. The proposed algorithm combines the strengths of multiple sensors and provides more precise location information. BIMFL uses the residual values of the measured data from various sensors as observations. Bayesian inversion effectively integrates prior information with observational information, both of which have their covariance matrices considered, hence determining the appropriate parameters of the fusing positioning system. Additionally, an update iterative process based on the Extended Kalman Filter (EKF) is adopted in BIMFL to refine positioning results. Simulation and experimental results have shown that BIMFL surpasses traditional single-sensor and EKF-based filtering methods.