underwater mapping

Underwater mapping 2025

Click on the octopus to return to the top of the page

17 - Cooperative SLAM Algorithm for Multi-AUV Underwater Exploration and Mapping

Author: Rongren Wu SLAM (Simultaneous Localisation and Mapping) is crucial for mapping unknown deep-sea areas. This paper suggests an AUV cluster SLAM algorithm to enhance mapping efficiency. It has three parts: (1) a multi-beam sonar image processing algorithm that filters dynamic points and redundant information; (2) SLAM using data from DVL, IMU, and DM based on a Rao-Blackwellised particle filter; and (3) an iUSBL system for cooperative positioning of AUVs. This algorithm significantly improves mapping efficiency and stabilizes SLAM performance by reducing errors from dynamic points.

16 - An overview of industrial image segmentation using deep learning models.

Authors: Guina Wang, Zhen Li, Guirong Weng, Yiyang Chen Image segmentation is important in AI and computer vision, used for tasks like industrial picking and video surveillance. This review covers traditional and deep learning segmentation algorithms, including their integration. It discusses various label frameworks and compares methods on benchmark datasets. Lastly, it addresses challenges and opportunities for future research.

01 - Remote Awareness of Image Quality for Multi-week Shore- launched AUV Surveys

Authors: Adrian Bodenmann, Daniel O.B. Jones, Alexander B. Phillips, Robert Templeton, Rashiid Sherif, Francesco Fanelli, Darryl Newborough, and Blair Thornton Visual seafloor imaging using autonomous underwater vehicles (AUVs) is now a common method for mapping and monitoring. This research creates a new way to assess image quality on-site during a 21-day AUV campaign. AUV operators received quality scores via satellite to improve data collection.

03 - Sound Absorption of the Water Column and Its Calibration for Multibeam Echosounder Backscattered Mapping in the East Sea of Korea

Authors: Seung-Uk Im, Cheong-Ah Lee, Moonsoo Lim, Changsoo Kim, and Dong-Guk Paeng This study uses Korea Oceanographic Data Center) datasets to analyze and visualize spatiotemporal variations in underwater sound absorption in the East Sea of Korea. The calibration process revealed discrepancies of up to 2.1 dB in backscatter intensity across survey lines, highlighting potential misrepresentation of legacy Multibeam echosounder (MBES) data due to site-specific absorption variability. This integrated approach enhances the reliability of legacy MBES data and provides valuable insights for marine resource management and environmental monitoring.

04 - Underwater DVL Optimization Network (UDON): A Learning- Based DVL Velocity Optimizing Method for Underwater Navigation

Authors: Feihu Zhang, Shaoping Zhao, Lu Li, and Chun Cao This paper proposes an end-to-end deep-learning approach to improve the accuracy of Doppler Velocity Log (DVL) velocity vector measurements for autonomous underwater vehicles (AUVs) that uses raw measurement data from an Inertial Measurement Unit (IMU) to assist DVL in velocity vector estimation and refining it towards the Global Positioning System (GPS). The research results show that the proposed method can achieve a maximum improvement of 69.26% in root mean square error and 78.62% in relative trajectory error in DVL vector predictions.

05 - Secrecy capacity maximization in autonomous underwater vehicle-enabled underwater acoustic sensor networks

Authors: Xufei Ding, Wen Tian, Xin Sun This study examines covert data collection in autonomous underwater vehicles (AUVs) for underwater acoustic sensor networks (UASNs). The researchers propose an efficient iterative optimization algorithm to optimize USN scheduling and AUV routes, aiming to maximize secrecy capacity under constraints such as AUV trajectory, USN scheduling, connection outage probability, and secrecy outage probability.

06 - Enhancing Physical Spatial Resolution of Synthetic Aperture Sonar Images Based on Convolutional Neural Network

Authors: Pan Xu, Dongbao Gao, Shui Yu, Guangming Li, Yun Zhao, and Guojun Xu This deep learning method is used to improve the physical spatial resolution of underwater sonar images using a proposed U-shaped end-to-end neural network, which includes down-sampling and up-sampling parts. This method reduces normalized loss on testing sets and improves the accuracy of predicted high-resolution images. This trained network can map high-resolution targets to impulse responses, providing a practical alternative for underwater target detection..

07 - Underwater Target Detection with High Accuracy and Speed Based on YOLOv10

Authors: Zhengliang Hu, Le Cheng, Shui Yu, Pan Xu, Peng Zhang, Rui Tian, and Jingqi Han This study presents an underwater target detection method based on YOLOv10 (You Only Look Once v10), which outperforms the YOLOv5 model in terms of detection accuracy and speed. The YOLOv10 model has a mAP50 (Average Precision at an Intersection over Union threshold of 50%) of 85.6% on the URPC 2020 (Underwater Robot Picking Contest 2020) dataset, a 1.2% improvement from YOLOv5, demonstrating its potential for precise and fast underwater target detection.

08 - Side-Scan Sonar Small Objects Detection Based on Improved YOLOv11

Authors: Chang Zou, Siquan Yu, Yankai Yu, Haitao Gu, and Xinlin Xu The study proposes an improved YOLOv11 (You Only Look Once v11) network called YOLOv11-SDC for underwater object detection using side-scan sonar. It includes a new Sparse Feature module, a Dilated Reparam Block, a Control Module, and a Content-Guided Attention Fusion module. The improved method outperforms previous versions in detection performance.

09 - A Deep Shrinkage Network for Direction-of-Arrival Estimation with Sparse Prior

Authors: Lei Zhou, Shihong Zhou, Yubo Qi, Lixin Wu, and Yannick Benezeth This study proposes a deep learning-based framework for estimating Direction-of-arrival (DOA) in underwater multipath signals. It uses sparse representation and a super-resolution deep shrinkage reconstruction network (SDSR-Net) to map the sparse representation of the covariance matrix to the DOA. Simulations and evaluations show that the proposed method significantly improves DOA estimation accuracy.

10 - An Improved YOLOv9s Algorithm for Underwater Object Detection

Authors: Shize Zhou, Long Wang, Zhuoqun Chen, Hao Zheng, Zhihui Lin, and Li He The YOLOv9s-SD (You Only Look Once 9s-SD) underwater target detection algorithm aims to improve underwater object detection performance by combining MobileNetV2's (MobileNetV2 is a convolutional neural network architecture designed for mobile and embedded vision applications) inverted residual structure with Simple Attention Module and Squeeze-and-Excitation Attention. Experiments show improvement in mean Average Precision, demonstrating better adaptability to intricate underwater environments.

11 - Advancing Seabed Bedform Mapping in the Kuznica Deep: Leveraging Multibeam Echosounders and Machine Learning for Enhanced Underwater Landscape Analysis

Authors: Lukasz Janowski This study focuses on the development of an automatic seabed mapping methodology using multibeam echosounders and machine learning, integrating various scientific fields to understand the underwater landscape, addressing sediment composition, backscatter intensity, and geomorphometric features. It emphasizes using fewer, carefully selected features for training models, improving accuracy and robustness.

12 - Real-Time Registration of 3D Underwater Sonar Scans

Authors: António Ferreira, José Almeida, Aníbal Matos, and Eduardo Silva This article presents a collection of generic matrix manipulation kernels for parallelized perception applications. It discusses a parallel implementation for the 3DupIC algorithm, a probabilistic scan matching method for sonar scan registration. Tests show the algorithm's real-time performance, enabling 3D sonar scan matching to be executed onboard the EVA AUV.

13 - Multi-Scale Feature Enhancement Method for Underwater Object Detection

Authors: Mengpan Li, Wenhao Liu, Changbin Shao, Bin Qin, Ali Tian, and Hualong Yu This paper presents a Multi-Scale Feature Enhancement (MSFE) method to address challenges in underwater object detection, aiming to achieve dual multi-scale information integration through the internal structural design of the basic C2F module (a module that converts convolutional outputs into structured features) in the Backbone network and the external global design of the feature pyramid network (FPN). The method uses a module to address fine-grained feature vanishing and injects a shallow feature branch to enhance information integration between inter-layer features.

14 - The Method for Storing a Seabed Photo Map of the During Surveys Conducted by an Autonomous Underwater Vehicle

Authors: Chang Liu, Vladimir Filaretov, Eduard Mursalimov, Alexander Timoshenko, and Alexander Zuev This paper presents a new method for creating a photographic map of the seabed using images captured by autonomous underwater vehicles' photo and video systems, stored in a mosaic format, allowing for automatic search and recognition of underwater objects. The authors emphasize the efficiency of the method, which forms the map directly during the vehicle's movement, using only the onboard computer's computing power. Its algorithm for saving the map in mosaic format, used in interactive geographic maps like Google Maps, reduces read and write operations, ensuring timely operation.

15 - AUV Online Path Planning Strategy Based on Sectorial Gridded Detection Area

Authors: Yang Liu, Jinxi Sun, Guojie Li, and Xiujun Xu This study proposes an online path planning strategy for an autonomous underwater vehicle using forward- looking sonar in unknown environments. It divides the sonar fan-shaped detection area into multiple grids and uses a comprehensive cost function for optimal heading. This strategy ensures safe and smooth navigation in obstructed areas and narrow passages.

20 - Lightweight Underwater Target Detection Algorithm Based on YOLOv8n

Authors: Dengke Song and Hua Huo The RDL-YOLO is a lightweight detection algorithm designed to tackle challenges in underwater target detection, including complex environments, image blurring, and high model parameter counts that incorporates optimizations. Experimental results show an average precision increase of 1.4% and 1.0%, respectively, while reducing parameter count and computational complexity by 19.3% and 14.8%.

21 - Persistent Localization of Autonomous Underwater Vehicles Using Visual Perception of Artifi cial Landmarks

Authors: Jongdae Jung, Hyun-Taek Choi, and Yeongjun Lee This paper proposes a visual landmarks-based simultaneous localization and mapping (SLAM) system for autonomous underwater vehicles (AUVs). The system uses two types of underwater landmarks observed using forward and downward-looking cameras. The information, along with incremental DR estimates, is integrated within the extended Kalman filter (EKF) SLAM approach.

22 - SDA-Mask R-CNN: An Advanced Seabed Feature Extraction Network for UUV

Authors: Yao Xiao, Dongchen Dai, Hongjian Wang, Chengfeng Li, and Shaozheng Song The paper introduces a novel framework for precise seabed terrain edge feature extraction from Side-Scan Sonar (SSS) images, enhancing Unmanned Underwater Vehicle (UUV) perception and navigation. This framework addresses challenges in underwater image analysis, including low segmentation accuracy and ambiguous edge delineation. It incorporates a Structural Synergistic Group-Attention Residual Network, a Depth-Weighted Hierarchical Fusion Network, and an Adaptive Synergistic Mask Optimization strategy.

23 - A Holistic High-Resolution Remote Sensing Approach for Mapping Coastal Geomorphology and Marine Habitats

Authors: Evagoras Evagorou, Thomas Hasiotis, Ivan Theophilos Petsimeris, Isavela N. Monioudi, Olympos P. Andreadis, Antonis Chatzipavlis, Demetris Christofi, Josephine Kountouri, Neophytos Stylianou, Christodoulos Mettas, Adonis Velegrakis, and Diofantos Hadjimitsis This research demonstrates how satellite, aerial, terrestrial, and marine remote sensing techniques can be integrated to produce accurate information, revealing significant elevation and shoreline changes over one year, highlighting coastal zone dynamics. This highlights the importance of incorporating information from all available sensors for complete geomorphological and marine habitat mapping for sustainable coastal management strategies.

24 - Model-Based AUV Path Planning Using Curriculum Learning and Deep Reinforcement Learning on a Simplifi ed Electronic Navigation Chart

Authors: Lukasz Marchel, Rafał Kot, Piotr Szymak, and Paweł Piskur This paper evaluates reinforcement learning (RL)-based methods for autonomous underwater vehicles (AUVs) trajectory planning. It incorporates a full motion model and proposes a reward function to elicit desired behaviors. The models are trained on simplified ENC maps and compared against five classical methods. The proposed solution enables efficient, collision-free route planning.

25 - Depth Estimation of an Underwater Moving Source Based on the Acoustic Interference Pattern Stream

Authors: Lintai Rong, Bo Lei, Tiantian Gu, and Zhaoyang He This study proposes an interference feature fusion (IFF) method for underwater maneuvering sources in deep oceans that extracts the principal interference mode of the VTR and constructs a depth ambiguity function. This method achieves long-term depth estimation and recognition of underwater sources without environmental information. It can suppress false peaks and reduce depth estimation errors by at least 47% compared to the generalized Fourier transform method.

26 - Underwater Side-Scan Sonar Target Detection: An Enhanced YOLOv11 Framework Integrating Attention Mechanisms and a Bi-Directional Feature Pyramid Network

Authors: Junhui Zhu, Houpu Li, Min Liu, Guojun Zhai, Shaofeng Bian, Ye Peng, and Lei Liu Underwater target detection is crucial for marine exploration, but faces challenges due to the complex underwater environment. The ABFP-YOLO model addresses these issues by integrating a bi-directional feature pyramid network and an attention module. It effectively recognizes targets of varying scales, especially small ones, in complex scenarios. Experiments on two datasets show the model achieves mean average precision scores of 0.988 and 0.866, indicating superior performance in complex underwater environments.

27 - Oscillatory Forward-Looking Sonar Based 3D Reconstruction Method for Autonomous Underwater Vehicle Obstacle Avoidance

Authors: Hui Zhi, Zhixin Zhou, Haiteng Wu, Zheng Chen, Shaohua Tian, Yujiong Zhang, and Yongwei Ruan This study proposes a cost-effective 3D reconstruction method for autonomous underwater vehicle inspection in 3D environments that uses an oscillatory forward-looking sonar with a pan-tilt mechanism, extending perception from a 2D plane to a 75-degree spatial range. The system integrates with the Ego-Planner path planning algorithm and nonlinear Model Predictive Control algorithm, achieving 100%, 60%, and 30% obstacle avoidance success rates under localization errors.

28 - Three-Dimensional Localization of Underwater Nodes Using Airborne Visible Light Beams

Authors: Jaeed Bin Saif, Mohamed Younis, and Fow-Sen Choa This paper proposes a novel localization method using visible light communication (VLC) from an airborne unit to locate underwater nodes, which uses VLC transmissions towards the water surface, encoded with GPS coordinates. Existing techniques assume depth measurements, but this method uses VLC signals to estimate the underwater node's 3D position. Extensive simulations validate the method's effectiveness and performance across various parameters.

29 - Underwater SLAM Meets Deep Learning: Challenges, Multi- Sensor Integration, and Future Directions

Authors: Mohamed Heshmat, Lyes Saad Saoud, Muayad Abujabal, Atif Sultan, Mahmoud Elmezain, Lakmal Seneviratne, and Irfan Hussain This survey analyzes the latest developments in deep learning-enhanced simultaneous localization and mapping (SLAM) for underwater applications, categorizing approaches based on methodologies, sensor dependencies, and integration with deep learning models, highlighting the benefits and limitations of existing techniques, key innovations, and unresolved challenges. It also introduces a novel classification framework for underwater SLAM based on its integration with underwater wireless sensor networks (UWSNs), enhancing localization, mapping, and real-time data sharing among AUVs.

30 - Underwater-Image Enhancement Based on Maximum Information-Channel Correction and Edge-Preserving Filtering

Authors: Wei Liu, Jingxuan Xu, Siying He, Yongzhen Chen, Xinyi Zhang, Hong Shu, and Ping Qi The authors propose an underwater-image enhancement (UIE) approach to address issues like color distortion and reduced contrast in underwater images. They use maximum information-channel compensation and edge- preserving filtering techniques to improve color quality. This approach involves enhancing the color-corrected image, generating globally contrast-enhanced images, and decomposing the enhanced image into low- frequency and high-frequency components.

31 - Task Allocation and Path Planning Method for Unmanned Underwater Vehicles

Authors: Feng Liu, Wei Xu, Zhiwen Feng, Changdong Yu, Xiao Liang, Qun Su, and Jian Gao This study proposes a task allocation and path planning method for Unmanned Underwater Vehicles (UUVs) to improve their autonomous navigation and adaptability in complex marine environments, using an Improved Grey Wolf Algorithm (IGWA) and a Circle chaotic mapping strategy to optimize task allocation and search efficiency. An algorithm termed GR-RRT is employed for path planning, ensuring obstacle avoidance and reducing redundant sampling. Experimental results show significant advantages in task completion rate and path optimization efficiency.

32 - YOLO-NeRFSLAM: underwater object detection for the visual NeRF-SLAM

Authors: Zhe Wang, Zhibin Yu, and Bing Zheng This paper proposes a new underwater Simultaneous Localization and Mapping (SLAM) framework that combines neural radiance fields (NeRF) with a dynamic masking module. The Marine Motion Fusion (MMF) strategy screens out dynamic objects, maintaining stable camera pose estimation and pixel-level dense reconstruction. Specialized loss functions cope with light attenuation and the dynamic nature of underwater scenes. Experimental results show that the method reduces localization drift, improves dense mapping accuracy, and achieves favorable runtime efficiency in multiple real underwater video datasets.

33 - Visual-Based Position Estimation for Underwater Vehicles Using Tightly Coupled Hybrid Constrained Approach

Authors: Tiedong Zhang, Shuoshuo Ding, Xun Yan, Yanze Lu, Dapeng Jiang, Xinjie Qiu, and Yu Lu This paper introduces a hybrid monocular visual Simultaneous Localization and Mapping (SLAM) system for unmanned underwater vehicles (UUVs), featuring a robust three-step tracking strategy. The system uses a feature-based method for rough pose estimation, a direct method for refining it, and a tightly coupled visual hybrid optimization method for stable tracking.

34 - Robust Forward-Looking Sonar-Image Mosaicking Without External Sensors for Autonomous Deep-Sea Mining

Authors: Xinran Liu, Jianmin Yang, Changyu Lu, Enhua Zhang, and Wenhao Xu This study introduces a robust Forward-Looking Sonar (FLS) image mosaicking framework for real-time environmental mapping in deep-sea mining vehicles (DSMV). This model integrates bidirectional cyclic consistency filtering with a soft-weighted feature refinement strategy and proposes a radial adaptive fusion algorithm for image fusion. The experimental results show high robustness and accuracy under real deep-sea conditions, supporting tasks like path planning, obstacle avoidance, and simultaneous localization and mapping, enabling reliable perceptual capabilities for intelligent underwater operations.

35 - Leveraging learned monocular depth prediction for pose estimation and mapping on unmanned underwater vehicles

Authors: Marco Job, David Botta, Victor Reijgwart, Luca Ebner, Andrej Studer, Roland Siegwart, and Eleni Kelasidi The paper introduces a framework that uses visual and acoustic sensor data to improve localization and mapping in complex underwater environments, particularly in fish farming. It enables net-relative pose estimation for Unmanned Underwater Vehicles (UUVs) and depth prediction using deep learning and sparse depth priors. This framework can generate detailed 3D maps in real- time, proving accurate for autonomous navigation and inspection.

36 - Direct Forward-Looking Sonar Odometry: A Two-Stage Odometry for Underwater Robot Localization

Authors: Wenhao Xu, Jianmin Yang, Jinghang Mao, Haining Lu, Changyu Lu, and Xinran Liu This paper proposes a lightweight front-end, forward- looking Sonar (FLS) odometry (DFLSO) for underwater robots, aiming to provide fast and accurate localization. It uses forward-looking sonar (FLS) images for pose estimation, extracting point clouds, and matching images. It also uses an image processing method to generate a 3- D point cloud, and a lightweight keyframe system to construct submaps, validated through simulation experiments and sea trials.

02 - Geomorphic time series reveals the constructive and destructive history of Havre caldera volcano, Kermadec arc

Authors: E. Spain, R. J. Carey, J. M. Whittaker, V. L. Lucieer, J. M. Fox, S. J. Watson, and F. Caratori Tontini This article examines geomorphic changes at the submarine Havre volcano over a 13-year period using a combination of bathymetric surveys and remotely operated vehicle (ROV) observations. By integrating various data sources, the study identifies four geomorphic groups and develops detailed maps that reveal volcanic features and processes. Notably, the study uncovers a significant underestimation of the scale of cryptodome emplacement related to the 2012 eruption and extends the timeline for volcanic product emplacement to a maximum of 3 years post-eruption. Overall, the research highlights ongoing seafloor modifications long after volcanic eruptions have occurred.

18 - Side-Scan Sonar Image Classification Based on Joint Image Deblurring–Denoising and Pre-Trained Feature Fusion Attention Network

Authors: Baolin Xie, Hongmei Zhang, and Weihan Wang This article discusses advancements in deep learning models for classifying side-scan sonar (SSS) images, focusing on addressing common challenges such as noise interference and limited dataset sizes. It introduces a novel classification model that combines joint image deblurring and denoising techniques with a pre-trained attention network. By leveraging advanced feature extraction methods and attention mechanisms, the model significantly improves classification accuracy and interpretability, achieving a notable accuracy rate of 96.80% on a dedicated dataset.

19 - A novel use of Stereo Baited Remote Underwater Video and Drop-Down Video for biodiversity and marine landscape mapping and prediction

Authors: Natasha L. Walker-Milne, Sophie Elliott, Peter J. Wright, & David M. Bailey This paper discusses the significance of benthic mapping in managing marine resources and restoring biodiversity, particularly through the analysis of patch- and landscape- level processes affecting marine species. The study utilizes Stereo Baited Remote Underwater Video (SBRUV) and Drop-Down Video (DDV) surveys conducted in the South Arran Marine Protected Area from 2013 to 2019. Through these surveys, a comprehensive dataset was generated to create detailed benthic maps that enhance understanding of substratum distribution and associated epibenthic biodiversity. The findings aim to improve management strategies for Marine Protected Areas by identifying areas of high biodiversity.

37 - Ship trajectory prediction via a transformer-based model by considering spatial-temporal dependency

Authors: Xinqiang Chen, Peiyang Wu, Yuzhen Wu, Loay Aboud, Octavian Postolache, Zichuang Wang The article discusses the increasing importance of ship trajectory prediction in global maritime trade, emphasizing the challenges posed by environmental complexities, multi-factor influences, and data quality issues in Automatic Identification System (AIS) data. It presents a novel ship trajectory prediction model based on the Crossformer architecture, which incorporates innovative components like Dimension-Segment-Wise embedding and a Hierarchical Encoder-Decoder structure. The model’s effectiveness is validated through experiments on public AIS datasets, showcasing its superior performance against traditional models.

38 - Submarine Topography Classification Using ConDenseNet with Label Smoothing Regularization

Authors: Jingyan Zhang, Kongwen Zhang, and Jiangtao Liu This article discusses the necessity of classifying submarine topography and geomorphology for various applications in marine resource exploitation and ocean engineering. It critiques traditional visual interpretation methods for their inefficiency and subjectivity while highlighting the challenges faced by current machine learning models when integrating multi-source data. The authors propose a new model, utilizing pruning-enhanced ConDenseNet with label smoothing regularization, which significantly improves classification accuracy and reduces computational complexity compared to leading models.

39 - Multi-Branch Towed Array System: Systematic Analysis of Modeling Methods, Environmental Responses and Mechanical Properties in Fracture Analysis

Authors: Jin Yan, Kefan Yang, Shengqing Zeng, Keqi Yang, Dapeng Zhang, and Keqiang Zhu The article discusses the critical role of multi-branch towed array systems in subsea information collection, emphasizing their significance in offshore construction and pipeline laying. It highlights the difficulties faced by these systems due to complex underwater conditions, including the risk of cable collisions and torsion interactions. Such challenges can lead to damaged towing systems and alterations in their formation, which are influenced by the specific sea state. The paper also draws a comparison between these towing systems and trawl nets to better understand their mechanical responses post-damage.

40 - Online Multi-AUV Trajectory Planning for Underwater Sweep Video Sensing in Unknown and Uneven Seafloor Environments

Authors: Talal S. Almuzaini, and Andrey V. Savkin This article discusses advancements in the navigation of Autonomous Underwater Vehicles (AUVs) for effective underwater sensing and monitoring. It presents a novel algorithm that tackles the challenges posed by uneven seafloor terrains and visibility constraints due to occlusions and camera limitations. The solution incorporates a trajectory planning algorithm that enhances coverage objectives while adhering to kinematic and safety constraints. The effectiveness of the proposed approach is demonstrated through simulations that show improved performance compared to traditional strategies.

43 - Adaptive Exposure Optimization for Underwater Optical Camera Communication via Multimodal Feature Learning and Real-to- Sim Channel Emulation

Authors: Jiongnan Lou, Xun Zhang, Haifei Shen, Yiqian Qian, Zhan Wang, Hongda Chen, Zefeng Wang, and Lianxin Hu Underwater Optical Camera Communication (UOCC) is a new method for secure data exchange in autonomous underwater vehicles. Its effectiveness relies on exposure time and International Standards Organization (ISO) sensitivity, affected by underwater conditions. Traditional systems struggle with changing conditions. A new framework combines imaging, environmental data, and a model for better prediction accuracy and real-time adaptation. This results in improved signal quality and potential for practical underwater communication use cases.

41 - Geostatistical uncertainty maps for real-world efficient AUV data collection

Authors: Ana F. Duarte, Lucrezia Bernacchi, Renato Mendes, João Borges de Sousa, and Leonardo Azevedo The article discusses a geostatistical modeling workflow designed to enhance the trajectory planning of Autonomous Underwater Vehicles (AUVs) for oceanographic surveys. It centers on two main objectives: predicting ocean temperature and spatial uncertainty, which informs optimized navigation paths, and facilitating the efficient assimilation of collected data into ocean models. An illustrative case study conducted off the coast of Western Portugal demonstrates the workflow’s efficacy, revealing improved data prediction capabilities and model reliability.

42 - RECAD: Retinex-Based Efficient Channel Attention with Dark Area Detection for Underwater Images Enhancement

Authors: Tianchi Zhang, Qiang Liu, Hongwei Qin, and Xing Liu The article focuses on improving visual target detection for Autonomous Underwater Vehicles (AUVs) by proposing a new image enhancement method named RECAD. This method addresses common problems found in weakly illuminated underwater images, such as blurring and color distortion. While existing deep learning methods struggle with feature extraction and high computational costs, RECAD utilizes Retinex theory and incorporates mechanisms for dark region detection and channel attention to enhance performance. Extensive testing shows that RECAD significantly outperforms existing models.

44 - Next-generation underwater localization: Artificial Intelligence- based and energy-aware approaches

Authors: Mainul Islam Chowdhury, Quoc Viet Phung, Iftekhar Ahmed, Walid K.Hasan, Daryoush Habibi This study highlights the problems associated with large propagation delays, the lack of GPS, node mobility, and limited acoustic link capacity that complicate underwater monitoring applications. Traditional localization methods’ limitations are discussed, particularly their sensitivity to noise and communication constraints. The article also emphasizes the importance of energy efficiency due to difficulties in replacing and recharging batteries in underwater environments. It provides a comprehensive review of how artificial intelligence (AI), including deep learning and machine learning, can enhance localization accuracy and robustness while integrating energy-saving strategies.

45 - A classification method of underwater target radiated noise signals based on enhanced images and convolutional neural networks

Authors: Yuan Muye, Lei Zhufeng, Dang Yinuo, and Shi Zanrong The article discusses the challenges of identifying underwater vehicles due to advancements in technology and the inadequacy of traditional classification methods based on signal characteristics. It presents a novel approach using multiple convolutional neural network (CNN) recognition methods enhanced by Gramian Angular Field (GAF) images. The paper details the process of converting underwater target noise signals into GAF images, which are then utilized to train various CNN models. The results indicate that the VGG-16 model demonstrated superior classification accuracy of 80.67%.

46 - RG-SAPF: A Scheme for Cooperative Escorting of Underwater Moving Target by Multi-AUV Formation Systems Based on Rigidity Graph and Safe Artificial Potential Field

Authors: Wen Pang, Daqi Zhu, Mingzhi Chen, & Wentao Xu This paper discusses how to escort a moving underwater target, like a human-occupied vehicle, using multiple autonomous underwater vehicles (AUVs) in difficult ocean settings. It introduces the “Reward-Guided State- Action Prediction and Filtering (RG-SAPF)” scheme for formation control and path planning, ensuring effective target escorting in complex environments.