CORTEXA
← Browse
crossrefJournal of Marine Science and Engineering2026-06-05Cited by 0

Multi-Source Sensor Fusion Localization Method for Autonomous Underwater Vehicles Based on Deep Learning

Xin Pan, Guoli Feng, Haiyan Zeng, Qunhong Tian

Autonomous Underwater Vehicles (AUVs) are increasingly used in deep-sea exploration, environmental monitoring, and marine engineering. Their operational safety and mission performance rely heavily on accurate and long-endurance underwater localization. However, both single-sensor localization methods and existing multi-sensor fusion approaches have inherent limitations, making it difficult to achieve high-precision localization during long-duration missions. To address this issue, this study develops a deep-learning-based multi-source sensor fusion framework for AUV localization. In the proposed framework, high-frequency data from the Inertial navigation system (INS) and Doppler velocity log (DVL) are used for continuous position propagation, while low-frequency absolute position observations from the Ultra-short baseline (USBL) system and Sonar are used to periodically correct the propagated results. Based on this framework, three instantiated models are developed using a Deep neural network (DNN), a Long short-term memory (LSTM) network, and a Bayesian semi-supervised mixed shallow-layer neural network (BSsMSLNN), respectively. Comparative experiments are conducted against the Extended Kalman filter (EKF) and Simultaneous localization and mapping system using Sonar, Visual, Inertial, and Depth sensor (SVIn2). The results show that the proposed framework effectively suppresses long-term error accumulation and significantly improves localization accuracy. Among the evaluated models, the BSsMSLNN-based method achieves the best performance in terms of trajectory fitting, root mean square error (RMSE), and coefficient of determination (R2). The proposed method provides a feasible solution for high-precision autonomous navigation of AUVs in GPS-denied environments.

View free PDFSource page

Related papers

openalexJournal of Marine Science and Engineering2026-07-24

Satellite-Derived Bathymetry for the Surveying of Coastal and Underwater Archaeological Sites: An Application to Ancient Asopos (Laconia, Greece)

Gerardo Diaz, Eleni Kolaiti

The use of Satellite-Derived Bathymetry (SDB) constitutes an efficient, cost-effective, time-saving, and scalable approach for generating high-resolution shallow-water bathymetry. In this context, SDB can be proven to be a valuable method for supporting bathymetric surveys of sub…

View free PDFSource page
crossrefJournal of Marine Science and Engineering2026-07-10

Machine-Learning-Assisted Prediction of Port-Flow Distribution and Multi-Objective Parametric Optimization for Navigation Lock Manifolds

Duo Xu, Zhonghua Li, Lingqin Mei, Tingqiang Xie

Navigation lock manifolds are key components of filling-and-emptying systems, and port-flow distribution affects chamber flow stability and filling efficiency. Under unsteady filling conditions, port-flow distribution is governed by discharge variation and manifold geometry, maki…

View free PDFSource page
crossrefJournal of Marine Science and Engineering2026-07-09

Toward Real-Time Shipwreck Detection for Autonomous Underwater Vehicles Using Deep Learning: A Model Evaluation Using High-Resolution Bathymetry Data

Agno Rubim de Assis, Thomas Guilment, Marco D’Emidio, Leonardo Macelloni

Autonomous Underwater Vehicles (AUVs) equipped with multibeam echosounders (MBESs) are deployed in oceans in expeditions worldwide to find shipwrecks, as they can survey the seafloor at the resolution required to identify such objects. Due to the severely constrained acoustic ban…

View free PDFSource page
crossrefJournal of Marine Science and Engineering2026-06-23

A Machine Learning Operations Framework for Self-Adaptive Anomaly Detection in Autonomous Surface Ships Under Data Drift

Minji Kim, Gwangho Yun, Hwasup Jang, Jaecheul Park

For stable operation of autonomous surface ships, real-time anomaly detection of engine conditions must be coupled with an operational framework that sustains model performance in dynamic maritime environments. This study proposes an autonomous maintenance system that combines a…

View free PDFSource page
crossrefJournal of Marine Science and Engineering2026-03-18Cited by 1

Satellite-Based Machine Learning for Temporal Assessment of Water Quality Parameter Prediction in a Coastal Shallow Lake

Anja Batina, Ljiljana Šerić, Andrija Krtalić, Ante Šiljeg

Satellite remote sensing increasingly supports water quality monitoring, yet the temporal transferability of machine learning (ML) models remains insufficiently tested, particularly in coastal shallow lakes subject to hydrological variability. This study evaluates the predictive…

View free PDFSource page
crossrefJournal of Marine Science and Engineering2026-02-25Cited by 1

Autonomous Navigation of an Unmanned Underwater Vehicle via Safe Reinforcement Learning and Active Disturbance Rejection Control

Qinze Chen, Yun Cheng, Yinlong Yuan, Liang Hua

A two-layer control framework for unmanned underwater vehicle (UUV) navigation is proposed, combining a lower-layer active disturbance rejection controller (ADRC) with an upper-layer safe reinforcement learning (RL) policy for obstacle-avoidance navigation. The lower layer, utili…

View free PDFSource page