IEEE Journal of Oceanic Engineering最新文献

{"title":"Toward Training-Free Underwater 3-D Object Detection From Sonar Point Clouds: A Comparison of Traditional and Deep Learning Approaches","authors":"Muhammad Salman Shaukat;Yannik Käckenmeister;Sebastian Bader;Thomas Kirste","doi":"10.1109/JOE.2026.3675841","DOIUrl":"https://doi.org/10.1109/JOE.2026.3675841","url":null,"abstract":"Underwater 3-D object detection remains one of the most challenging frontiers in computer vision, where traditional approaches struggle with the harsh acoustic environment and scarcity of training data. While deep learning has revolutionized terrestrial 3-D detection, its application underwater faces a critical bottleneck: obtaining sufficient annotated sonar data is prohibitively expensive and logistically complex, often requiring specialized vessels, expert surveyors, and favorable weather conditions. This work addresses a fundamental question: Can we achieve reliable underwater 3-D object detection without real-world training data? We tackle this challenge by developing and comparing two paradigms for training-free detection of artificial structures in multibeam echo-sounder point clouds. Our dual approach combines a physics-based sonar simulation pipeline that generates synthetic training data for state-of-the-art neural networks, with a robust model-based template matching system that leverages geometric priors of target objects. Evaluation on real bathymetry surveys from the Baltic Sea reveals important insights: neural networks trained on synthetic data suffer severe domain shift, with performance dropping by approximately 70% from simulated to real sonar data (30% mAP at IoU = 0.5). Our template matching approach proves far more stable, experiencing an approximate 30% drop from simulation to reality and maintaining 68% mAP on real data without requiring any training, demonstrating remarkable robustness to acoustic noise and environmental variations. Our findings challenge conventional wisdom about data-hungry deep learning in underwater domains and establish an initial comparative benchmark for training-free underwater 3D detection on large-area seafloor surveys. This work opens new possibilities for autonomous underwater vehicle navigation, marine archaeology, and offshore infrastructure monitoring in data-scarce environments where traditional machine learning approaches fail.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"2284-2297"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Obtaining a Reverberant Tank’s Transfer Function and Its Application to the Calibration of Acoustic Transducers and Recorders","authors":"Pedro Mendes Diniz;Caio César Marques Pereira de Alcântara;Stilson Veras Cardoso;Rafael Bruno de Jesus Trindade;Rodrigo Soares de Freitas;Fernando Luiz de Magalhães;Orlando de Jesus Ribeiro Afonso","doi":"10.1109/JOE.2026.3674035","DOIUrl":"https://doi.org/10.1109/JOE.2026.3674035","url":null,"abstract":"This article describes a method for the experimental determination of the reverberant field transfer function of an acoustic test tank in the frequency domain. Once obtained, the transfer function enables the calibration of transducers and marine recorders in a reverberant field. The method relies on a room acoustics framework for mathematically separating the direct and reverberant fields in a point inside the tank, but employs a transducer-centric model that dispenses with extensive characterization of the tank itself. A calibration experiment in the 1–24 kHz band of an omnidirectional SX147 transducer is conducted in the reverberant field, resulting in mean absolute errors of 0.78 and 1.00 dB, respectively for the receive and transmit voltage responses, relative to the free-field calibration. The expanded uncertainty (coverage factor of 2) in the calibration arising from the spatial averaging of the transfer function ranges from 4.63% to 8.57% . The model is extended to include a transfer function that models the effect of an acoustic recorder’s body on the field near the hydrophone, presenting a feasible method for obtaining the sensitivity of the recorder. An experiment shows that the free-field sensitivity of the standalone hydrophone can be recovered after compensating the measurements for the tank’s and recorder’s transfer functions with absolute errors smaller than 0.92 dB. The expanded uncertainty due to spatial averaging of 20 measurements in determining the recorder’s transfer function ranges from 6.31% to 13.23% .","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"2409-2420"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452072","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Real-Time Underwater Visual Target Sensing System With Multitasking via Edge-Cloud Collaboration","authors":"Wenli Wu;Xiaoning Huang;Bohuan Xue;Xiaoyu Tang","doi":"10.1109/JOE.2026.3678825","DOIUrl":"https://doi.org/10.1109/JOE.2026.3678825","url":null,"abstract":"Underwater object detection (UOD) is significant in marine life protection and resource development. Underwater images frequently exhibit degraded characteristics, such as severe fogging, color distortion, and edge distortion. These challenges result in poor performance of general object detectors in UOD tasks. To address these challenges, this study proposes a novel underwater visual target sensing system (UVTS<inline-formula><tex-math>${}^{2}$</tex-math></inline-formula>). The system includes a lightweight underwater image enhancement-detection network and an edge-cloud collaboration (ECC) mode. Specifically, our proposed enhancement network, lightweight-and-effective enhancement network (LE<inline-formula><tex-math>${}^{2}$</tex-math></inline-formula>Net), takes efficiency and accuracy as the design indicators and achieves real-time image enhancement through a redesigned wide-range channel attention module. The proposed detection network, multilayer supervision network (MLSNet), introduces the mechanism of deep supervision. It redesigns the transmission path of supervisory information to improve the overall fitting ability of the model. At the same time, the redesigned ECC framework fully incorporates the specific characteristics of visual multitasking. We propose and experimentally validate the UVTS<inline-formula><tex-math>${}^{2}$</tex-math></inline-formula>. The results show that the proposed enhancement network LE<inline-formula><tex-math>${}^{2}$</tex-math></inline-formula>Net effectively improves the quality of underwater images using only 300 K parameters and significantly affects the performance in UOD and other tasks. Compared with the baseline, MLSNet improves detection accuracy by 1.5% on the RUOD data set while reducing 11 million parameters and achieves good results on the COCO data set, verifying the generalization of the model. In addition, the proposed ECC deployment scheme notably enhances the rate of inference frames.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"2430-2443"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452103","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"BlueME: Robust Underwater Robot-to-Robot Communication Using Compact Magnetoelectric Antennas","authors":"Mehron Talebi;Sultan Mahmud;Adam Khalifa;Md Jahidul Islam","doi":"10.1109/JOE.2026.3675822","DOIUrl":"https://doi.org/10.1109/JOE.2026.3675822","url":null,"abstract":"We present the design, development, and experimental validation of <italic>BlueME</i>, a compact magnetoelectric (ME) antenna array system for underwater robot-to-robot communication. BlueME employs ME antennas operating at their natural mechanical resonance frequency to efficiently transmit and receive very-low and low-frequency (VLF/LF) electromagnetic signals underwater. We detail the system's design, simulation, fabrication, and integration onto low-power embedded platforms, emphasizing portability and scalability. For performance evaluation, we deployed BlueME on mobile robot platforms in freshwater (lake) and saltwater (ocean) trials. Our tests demonstrate reliable signal transmission and detection using the BlueME antenna system at distances beyond 700 m while consuming less than 10 W of power. Ocean trials demonstrate that the system operates effectively under challenging conditions, such as turbidity, obstacles, and multipath interference, which are factors that typically degrade acoustic and optical methods. Our analysis also examines the impact of complete submersion on system performance and identifies key deployment considerations. This work represents the first practical underwater deployment of ME antennas outside the laboratory and implements the largest VLF/LF ME array system to date. BlueME demonstrates significant potential for marine robotics and automation in multirobot cooperative systems and remote sensor networks.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"1718-1733"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452113","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dual-Critic Proximal Policy Optimization for Trajectory Tracking of Autonomous Underwater Vehicle Under Time-Varying Dynamics","authors":"Ye Yuan;Chaoyang Chen;Dan Yang;Weidong Zhang;Yaonan Wang","doi":"10.1109/JOE.2026.3692007","DOIUrl":"https://doi.org/10.1109/JOE.2026.3692007","url":null,"abstract":"The autonomous underwater vehicle (AUV) serves as an ideal platform for replacing human operators in high-risk underwater operations, where high-precision trajectory tracking control is crucial for ensuring mission effectiveness. However, the time-varying dynamics caused by the complexity of the underwater environment present significant challenges to achieving high-precision trajectory tracking for AUVs. To address this challenge, a model-assisted deep reinforcement learning-based trajectory tracking method for an AUV with time-varying dynamics is presented in this article. Specifically, a novel state space is proposed, which replaces the absolute position of the agent with the relative position. This representation provides a more comprehensive description of the relationship between the AUV and the reference trajectory. The policy trained based on this state space can be easily transferred and applied to a different type of reference trajectory. Furthermore, by decomposing the reward function into three components, the sparse reward issue is effectively mitigated. The smoothness term within the reward function effectively suppresses abrupt changes in control commands, ensuring that the AUV maintains smooth navigation even under time-varying dynamics. Finally, an improved proximal policy optimization algorithm is proposed. The algorithm includes two critics that separately evaluate the value functions of actions under time-varying and static dynamics, and adaptively adjust their weights based on the intensity of the time-varying dynamics, effectively enhancing the policy’s disturbance rejection capability and robustness. Experimental results demonstrate that the proposed method maintains high-precision trajectory tracking across various time-varying dynamics, significantly outperforming other algorithms and validating its effectiveness.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"1899-1912"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design and Validation of a Wave-Inversion-Compensated Path-Following Controller for Unmanned Surface Vehicle: Simulation and Experiment","authors":"Dongdong Mu;Guilin Xue;Yongsheng Zhao;Yukun Hu;Xiao Liang;Yunsheng Fan","doi":"10.1109/JOE.2026.3697327","DOIUrl":"https://doi.org/10.1109/JOE.2026.3697327","url":null,"abstract":"To compensate for wave disturbances encountered by an uncrewed surface vehicle (USV) during navigation, the application of feedforward compensation using the wave information estimated from waves is a good strategy. However, the computational complexity of wave estimation does not meet the real-time requirements of the control system. To reduce the complexity of ship motion response-based wave estimation, and effectively integrate wave estimation into the USV control system while balancing real-time performance and estimation accuracy, a wave inversion strategy based on USV motion time-series data is proposed in this article. The strategy combines the particle swarm optimization (PSO) algorithm with the sequential quadratic programming (SQP) method to achieve a favorable tradeoff between the real-time performance and accuracy of wave estimation, and is further embedded into the design of the USV path-following controller. The method compensates for most of the wave disturbance at the physical level and improves the control effect of the USV. First, using the time-history data of the pitch under low-sea conditions, a hybrid PSO-SQP algorithm is constructed to estimate wave parameters and response amplitude operators, leading to the establishment of an estimated wave spectrum. The effectiveness of the estimation method is validated through simulations with synthetic data. Second, an estimated wave disturbance model is developed and integrated into the control system, where the extended state observer-line of sight guidance law and the backstepping method are used for controller design. To address external disturbances beyond wave disturbance, an radial basis function neural network is adopted to approximate and compensate for errors. Finally, the proposed approach is further validated through simulation analysis and field experiments. In the experimental phase, real pitch data from the USV are used to estimate ocean waves, and a virtual navigator strategy is implemented. Path-following control tests are conducted on two different USV experimental platforms, achieving favorable control results and demonstrating that the method is both applicable and feasible across different platforms under sea states of level 2 and 3.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"1833-1861"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Bernoulli Particle Filter for Target Tracking in a Radar Detection-Occluded Environment","authors":"Shuai Sun;Guchong Li;Hongguang Lyu","doi":"10.1109/JOE.2026.3674016","DOIUrl":"https://doi.org/10.1109/JOE.2026.3674016","url":null,"abstract":"In maritime environments, target detection using radar signals often suffers from occlusion by islands or other vessels, particularly in where the radar platform is a small autonomous surface vehicle (ASV) with a civil marine radar with a low-height viewing angle. Low antenna height on a ASV platform can significantly reduce the spatial coverage of the radar beam as a result of signal occlusion. Once a high value target is occluded, continuous tracking of it may be interrupted because of the lack of radar measurements. In the literature, target occlusion is often ignored or only partly addressed for a specific target tracking scenario. This article proposes to use Bernoulli random finite sets to model the appearance and disappearance of a target in a radar detection occluded environment, and a particle-based implementation to produce an algorithm with environmental occlusion awareness. More specifically, we first segment the environment using uniform grid cells and then check the line-of-sight condition between each particle and the radar platform for occlusion. This method is tested using both simulated and real radar data. The results demonstrate that our proposed solution outperforms existing trackers that ignore target occlusion.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"2109-2121"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452121","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Energy-Optimal Trajectory Planning for Autonomous Underwater Vehicles Using Pontryagin’s Maximum Principle","authors":"Raymond Young;F. Hunter Akins;Sophia T. Merrifield","doi":"10.1109/JOE.2026.3692006","DOIUrl":"https://doi.org/10.1109/JOE.2026.3692006","url":null,"abstract":"Pontryagin’s Maximum Principle, a classical tool from optimal control theory, is used to compute energy-optimal trajectories for an autonomous underwater vehicle (AUV) navigating in deterministic, complex flows. This approach reformulates the optimal control problem as a two-point boundary value problem (TPBVP) for ordinary differential equations (ODEs) and offers computational benefits compared to alternative solutions which require the solution of partial differential equations (PDEs). This study introduces a numerical framework using pure shooting to solve the associated TPBVP and demonstrates its efficacy for trajectory optimization through analytic flows and current fields from high-resolution ocean simulations. The method is tested on two case studies: first, a multiday transit through the Southern California Bight, where ocean currents are dominated by mesoscale eddies and tidal variability, and energy-optimal planning predicts over 20% of savings relative to a straight-line transit, and second, a multihour transit off the coast of Ram Head, St. John, U.S. Virgin Islands, where the flow is dominated by energetic tidal vortices and energy consumption varies by up to 50% within a 6 h launch window. In both cases, the tradeoff between time and energy is analyzed for a wide range of AUVs that are parameterized by their nominal speed through water, between 0.25 and 2.0 m/s.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"1747-1766"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11556408","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Call for Papers: Special Issue on the IEEE UT2027 Symposium","authors":"","doi":"10.1109/JOE.2026.3711922","DOIUrl":"https://doi.org/10.1109/JOE.2026.3711922","url":null,"abstract":"","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"2555-2556"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11611563","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148452156","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Ensemble of Vision Transformers for Sea State Classification","authors":"Marina Ivasic-Kos;Domagoj Palinic;Miran Pobar","doi":"10.1109/JOE.2026.3674388","DOIUrl":"https://doi.org/10.1109/JOE.2026.3674388","url":null,"abstract":"This study presents an innovative method for real-time sea state classification that utilizes an ensemble-based computer vision framework. Multiple pretrained submodels, each individually fine-tuned on specialized data sets, are integrated to recognize eight classes on the Beaufort scale. A comprehensive multimodal data set was assembled, combining visual information with wind speed data from anemometers and further optimized through preprocessing steps. Standalone models based on cutting-edge architectures, such as the data efficient vision transformer and EfficientNet deep convolutional network, were adapted specifically for sea state recognition. Initial results yielded F1 scores of 66% and 65%, respectively, for these models. To boost overall accuracy, a weighted majority voting and precision-driven ensemble strategy was adopted. The unimodal ensemble achieved 73% F1 score, while the multimodal ensemble, leveraging both image and wind data, reached an F1 score of 97%. For a reduced six-class scenario, where noncritical classes were grouped, the unimodal ensemble reached an impressive F1 score of 94%. These outcomes illustrate the potential of this ensemble approach to significantly advance maritime safety and operational efficiency in sea state monitoring.","PeriodicalId":13191,"journal":{"name":"IEEE Journal of Oceanic Engineering","volume":"51 3","pages":"2122-2136"},"PeriodicalIF":5.6,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11510753","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148451885","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}