Embedded Multi-Rate Sensor Fusion for Underwater Navigation: Experimental Assessment in Confined Environments

Underwater navigation in the absence of Global Navigation Satellite System signals remains challenging due to limitations in underwater sensing and communication. This paper presents an embedded multi-rate sensor fusion architecture for underwater position estimation, designed as a modular payload for generic underwater platforms. The proposed system integrates an inertial measurement unit, a depth sensor, and range-only Ultra-Short Baseline (USBL) acoustic positioning within an Extended Kalman Filter framework handling asynchronous sensor updates. The architecture is implemented on an embedded computing platform and experimentally evaluated through two test campaigns under static and dynamic conditions, conducted in a confined pool environment, where multipath propagation and unfavorable acoustic geometry degrade the quality of USBL measurements. Experimental results show stable convergence of the proposed estimator when consistent acoustic updates are available, highlighting the limitations of range-only USBL measurements in dynamic scenarios.