Studies of ROV concepts 2025 - Part C
Back to the menu Next page Next page BUTTON TEXT Previous page
Click on the octopus to return to the top of the page
02 - Translation-based multimodal learning: a survey
Authors: Zhengyi Lu, Yunhong Liao, & Jia Li The article provides a comprehensive survey of translation- based multimodal learning, focusing on two main paradigms: end-to-end translation and representation- level translation. End-to-end methods facilitate direct mappings between modalities using various advanced architectures, while representation-level methods align multimodal signals within a common embedding space, enhancing robustness against data noise. Highlighting insights from over forty benchmark studies, the article discusses two notable models—the xDSBMIT and TransTrans—demonstrating their effectiveness in multimodal translation tasks. Lastly, it identifies ongoing challenges and future directions for developing versatile and ethically sound multimodal systems.
03 - Powering Underwater Robotics Sensor Networks Through Ocean Energy Harvesting and Wireless Power Transfer Methods: Systematic Review
Authors: Sverrir Jan Nordfjord, Saemundur E. Thorsteinsson, and Kristinn Andersen The article discusses the rising global demand for innovative underwater applications, highlighting the energy challenges faced by underwater devices used in scientific research, commercial ventures, and defense operations. It focuses on two primary methods for powering these devices: energy harvesting from natural environmental sources and wireless power transfer (WPT). While energy harvesting is dependent on specific environmental conditions, WPT faces efficiency issues over distance. The paper concludes that further research is necessary to develop hybrid solutions that can effectively utilize multiple power sources.
04 - Underwater Acoustic Integrated Sensing and Communication: A Spatio-Temporal Freshness for Intelligent Resource Prioritization
Authors: Ananya Hazarika, & Mehdi Rahmati This article introduces an integrated sensing and communication (ISAC) framework specifically designed for underwater environments, which uses a new Spatio- Temporal Information-Theoretic Freshness metric. This approach shifts the focus from merely minimizing delays to maximizing the value of transmitted information. The framework includes intelligent strategies for power allocation, beamforming, and cooperative sensing, supported by multi-agent reinforcement learning for enhanced resource use. The proposed methods demonstrate significant performance improvements in energy efficiency and data relevance in realistic underwater scenarios.
05 - Influence of Marine Environmental Factors on Characteristics of Composite Magnetic Field of Underwater Vehicles
Authors: Honglei Wang, Xinyu Dong, and Yixin Yang This research study explores the spatiotemporal characteristics, attenuation laws, and influencing mechanisms of these fields. The study integrates three distinct types of magnetic fields: the magnetic anomaly field created by the ferromagnetic vehicle, the wake magnetic field from its motion, and the ocean wave magnetic field influenced by wind-driven waves. Findings reveal that vehicle motion parameters affect the magnetic anomaly and wake fields, while wind conditions significantly impact the ocean wave magnetic field. Overall, this work improves the understanding of how marine environmental factors and vehicle dynamics contribute to the magnetic fields during underwater navigation.
06 - Inductive Wireless Power Transfer for Autonomous Underwater Vehicles: A Comprehensive Review of Technological Advances and Challenges
Authors: Han Xu, Rong Zheng, & Bo Yang, and Wei Ning The article discusses the advancements in underwater inductive wireless power transfer (UIWPT) as a solution to enhance the endurance and efficiency of autonomous underwater vehicles (AUVs) by providing efficient energy replenishment. It systematically reviews the architecture of UIWPT systems, evaluates power loss mechanisms and optimization strategies, and summarizes recent research developments in relevant technologies. The discussion extends to environmental factors affecting UIWPT integration with AUV platforms and concludes with future trends and challenges in deploying reliable underwater power systems.
07 - Visual Signal Recognition with ResNet50V2 for Autonomous ROV Navigation in Underwater Environments
Authors: Cristian H. Sánchez-Saquín, Alejandro Gómez- Hernández, Tomás Salgado-Jiménez, Juan M. Barrera Fernández, Leonardo Barriga-Rodríguez, and Alfonso Gómez-Espinosa This article details the development and assessment of AquaSignalNet, a deep learning system designed for recognizing underwater visual commands, thus facilitating autonomous navigation for Remotely Operated Vehicles (ROVs). Utilizing a ResNet50 V2 architecture and a custom dataset, UVSRD, the system was trained on 33,800 labeled images representing various gesture classes. Deployed on a Raspberry Pi 4, AquaSignalNet demonstrated high success rates in navigating predefined paths, showing promise as a markerless solution for underwater tasks.
08 - An Underwater Salvage Robot for Retrieving Foreign Objects in Nuclear Reactor Pools
Authors: Ming Zhong, Zihan Gao, Zhengxiong Mao, Ruifei Lyu, and Yaxin Liu The article presents the development of an underwater salvage robot designed to locate and retrieve foreign objects in nuclear reactor pools. The robot features an ROV platform combined with a three-degree-of-freedom Delta robotic arm and utilizes advanced technologies for localization and object detection. Experimental results demonstrate the robot’s high effectiveness, achieving notable identification and retrieval rates of foreign objects.
09 - Deep learning methods for 3D tracking of fish in challenging underwater conditions for future perception in autonomous underwater vehicles
Authors: Martin Føre, Emilia May O’Brien, and Eleni Kelasidi Unmanned underwater vehicles (UUVs) are increasingly employed in fish farming to perform tasks unsuitable for human workers. However, there is concern that these vehicles could stress farmed fish by causing them to flee from intrusive objects. To address this issue, researchers have developed automated systems that can assess the distance between fish and the UUVs in real-time using modern camera technology and deep learning methods. This involves four innovative pipelines that facilitate the measurement of distances, which were validated against sonar data. The best-performing methods were integrated into the Robot Operating System (ROS2) to improve fish welfare and enhance control over UUVs.
10 - A Novel Cooperative Navigation Algorithm Based on Factor Graph and Lie Group for AUVs
Authors: Jiapeng Liu, Xiaodong Bu, & Chao Wu The article introduces a new cooperative navigation algorithm for multiple Autonomous Underwater Vehicles (AUVs) that is versatile and applicable to various configurations. Unlike traditional algorithms designed for specific forms (like parallel or leader-slave), this method utilizes factor graphs and Lie group theory to enhance AUV localization capabilities. By representing AUV motion states in the two-dimensional special Euclidean group and employing nonlinear optimization techniques, the new algorithm demonstrates substantial improvements in localization accuracy across different operational setups, significantly outperforming conventional Extended Kalman Filter (EKF) methods.
11 - A CFD-Based Surrogate for Pump–Jet AUV Maneuvering
Authors: Younhee Kwon, Dong-Hwan Kim, Jeonghwa Seo, and Hyun Chung This article presents a surrogate maneuvering model developed to predict the performance of autonomous underwater vehicles (AUVs) equipped with pump-jet propulsion. Traditional methods relying on high-fidelity computational fluid dynamics (CFD) are computationally expensive. The surrogate model provides comparable accuracy while significantly reducing costs, utilizing results from unsteady Reynolds-averaged Navier-Stokes equations and employing a Taylor-expansion-based framework. It improves the representation of propulsion and rudder forces and has been validated against high-fidelity simulations, demonstrating its ability to accurately forecast kinematic responses, forces, and moments.
12 - Deep Learning-Assisted ES-EKF for Surface AUV Navigation with SINS/GPS/DVL Integration
Authors: Yuanbo Yang, Bo Xu, Baodong Ye, & Feimo Li The article introduces a new cooperative navigation algorithm for multiple Autonomous Underwater Vehicles (AUVs) that is versatile and applicable to various configurations. Unlike traditional algorithms designed for specific forms (like parallel or leader-slave), this method utilizes factor graphs and Lie group theory to enhance AUV localization capabilities. By representing AUV motion states in the two-dimensional special Euclidean group and employing nonlinear optimization techniques, the new algorithm demonstrates substantial improvements in localization accuracy across different operational setups, significantly outperforming conventional Extended Kalman Filter (EKF) methods.
13 - Multi-AUV Cooperative Search for Moving Targets Based on Multi-Agent Reinforcement Learning
Authors: Le Li, Ruiqi An, Zhaozhi Guo, & Jian Gao The paper presents a novel convolutional multi-agent deep deterministic policy gradient method, termed PER- CMADDPG, designed for multi-autonomous underwater vehicle (AUV) cooperative search of moving targets. It establishes a mathematical model comprising environment, AUV, and information update systems. The PER-CMADDPG method innovates on the MADDPG framework by integrating convolutional neural networks (CNNs) for improved spatial awareness and employing a prioritized experience replay (PER) to enhance learning efficiency. Simulation results illustrate that this method accelerates convergence and achieves greater rewards, while also analyzing the effects of various operational parameters
14 - Robust Underwater Docking Visual Guidance and Positioning Method Based on a Cage-Type Dual-Layer Guiding Light Array
Authors: Ziyue Wang, Xingqun Zhou, Yi Yang, Zhiqiang Hu, Qingbo Wei, Chuanzhi Fan, Quan Zheng, Zhichao Wang, and Zhiyu Liao This paper introduces a cage-type dual-layer guiding light array that enhances visual localization as the Autonomous Underwater Vehicle (AUV) approaches the docking station. It outlines a dynamic localization algorithm designed to differentiate beacon appearance at various docking stages, applying advanced techniques like particle swarm optimization and a robust filtering strategy. The proposed method demonstrates high success rates in beacon matching, even under extreme conditions, facilitating reliable and continuous docking guidance for AUVs.
15 - Marine-Inspired Multimodal Sensor Fusion and Neuromorphic Processing for Autonomous Navigation in Unstructured Subaquatic Environments
Authors: Chandan Sheikder, Weimin Zhang, Xiaopeng Chen, Fangxing Li, Yichang Liu, Zhengqing Zuo, Xiaohai He, and Xinyan Tan This article proposes a novel neuromorphic framework inspired by marine species’ navigation strategies, emphasizing the co-design of advanced sensors and event-based neuromorphic processors. This framework aims to improve navigational robustness and energy efficiency by mimicking the multisensory fusion and processing capabilities found in marine life, suggesting significant advancements for applications like deep-sea exploration and underwater infrastructure inspection.
22 - Physics-Informed Dynamics Modeling: Accurate Long-Term Prediction of Underwater Vehicles with Hamiltonian Neural ODEs
Authors: Xiang Jin, Zeyu Lyu, Jiayi Liu, and Yu Lu This document introduces a physics-informed framework that models the dynamics of an Autonomous Underwater Vehicle (AUV) through the discovery of its energy landscape, integrating Port-Hamiltonian mechanics into a neural ordinary differential equation (NODE) structure. This approach allows for rigorous identification of the AUV’s Hamiltonian and associated physical matrices from observational data, achieving superior accuracy and consistency compared to traditional black-box deep learning models, especially during complex maneuvers.
23 - Distributed Adaptive Fault-Tolerant Formation Control for Heterogeneous USV-AUV Swarms Based on Dynamic Event Triggering
Authors: Haitao Wang, Hanyi Wang, & Xuan Guo This article discusses a novel approach to cooperative formation control for heterogeneous unmanned systems, specifically Unmanned Surface Vehicles (USVs) and Autonomous Underwater Vehicles (AUVs), addressing challenges such as actuator faults, varying communication topologies, and communication delays. A unified dynamic model utilizing the Euler–Lagrange framework is developed, along with a Distributed Adaptive Fault- Tolerant Control (DAFTC) that incorporates a Dynamic Event-Triggered Mechanism (DETM) to optimize communication efficiency while maintaining robustness through online fault compensation.
01 - Deformable USV and Lightweight ROV Collaboration for Underwater Object Detection in Complex Harbor Environments: From Acoustic Survey to Optical Verification
Authors: Yonghang Li, Mingming Wen, Peng Wan, Zelin Mu, Dongqiang Wu, Jiale Chen, Haoyi Zhou, Shi Zhang, and Huiqiang Yao This paper presents a USV-ROV collaborative system for efficient underwater inspection and object disposal in complex harbor environments. The deformable USV, equipped with side-scan sonar (SSS) and multibeam echo sounder (MBES), enables rapid large-area surveying and high-resolution 3D seabed mapping; it accurately localized and preliminarily identified simulated objects with ≤2.2 m positioning deviation. The lightweight ROV, fitted with an optical camera, forward-looking sonar, and a manipulator, performed close-range verification and disposal, confirming object identity and location with ≤3.2 m deviation from SSS estimates.
26 - Research and Development of a Autonomous Underwater Vehicle
Authors: Mustafa Cenik, Mehmet Talha Yilmaz, Umut Can Senuzar, Ömer Eroglu, and Orçin Uygun This study outlines the design of a fully autonomous unmanned underwater vehicle (UUV) aimed at enhancing automation in aquaculture, aquatic scientific research, and national defense. The prototype uses a Raspberry Pi 5 for AI-driven image processing (via Raspberry Pi Camera V2) and real-time sensor data handling—including motor control, orientation, depth, and temperature. Autonomous navigation and decision- making are implemented through embedded Python and AI algorithms. Three-axis movement is enabled by a Pixhawk PX4 flight controller and six motors.
27 - Digital twin-driven swarm of autonomous underwater vehicles for marine exploration
Authors: Jing Yan, Tianyi Zhang, Xinping Guan, Xian Yang & Cailian Chen This article proposes a digital twin (DT)-driven swarm control framework for autonomous underwater vehicles (AUVs) to address two key challenges in marine exploration: high communication energy consumption and limited environmental perception. It constructs DT models for each AUV by fusing dynamics and environmental data; employs a parameter estimator to predict underwater flow fields; designs an energy-efficient swarm networking protocol; introduces an integral reinforcement learning (IRL)-based controller for coordinated virtual and physical AUV operation; and applies a virtual-real error optimization algorithm to enhance model fidelity.
28 - Disturbance-Rejection Control Strategies and Algorithms for Autonomous Underwater Vehicles and Unmanned Aerial Vehicles: A Cross-Domain Survey
Author: Yanzhou Wang This paper reviews disturbance-rejection control strategies for autonomous underwater vehicles (AUVs) and unmanned aerial vehicles (UAVs) operating in fluid- disturbed environments. It contrasts the two platforms' dynamic structures, medium properties, and disturbance types—highlighting differing requirements for low- frequency and high-frequency disturbances. The study examines key control methods, including sliding mode control (SMC), model predictive control (MPC), active disturbance rejection control (ADRC), and intelligent approaches like deep reinforcement learning and neural networks, assessing their robustness and applicability. Comparative analyses and guidelines are provided to aid practical control strategy design.
29 - The Provision of Physical Protection of information During the Transmission of Commands to a Group of UAVs Using Fiber Optic Communication Within the Group
Author: Dina Shaltykova, Aruzhan Kadyrzhan, Yelizaveta Vitulyova, and Ibragim Suleimenov This paper presents a novel method for the precise localization of remote radio-signal sources using a formation of unmanned aerial vehicles (UAVs). By studying the asymptotic intersections of these hyperbolas, the method ensures unique determination of the source position, even in the presence of multiple intersection points. Theoretical analysis confirms that the correct intersection point is located at a significantly larger distance from the UAV formation center compared to spurious intersections, providing a rigorous criterion for resolving localization ambiguity.
16 - Trust your MUM: Trust as a Pillar of User Acceptance for the Autonomous Modifiable Underwater Mothership (MUM)
Authors: A Schmitz and H Braun The MUM project develops a modular, fuel cell-powered autonomous underwater vehicle for extended missions in remote, high-risk areas like under Arctic ice, where operator trust is essential for accepting control transfer. Using mixed qualitative and conceptual methods, the paper identifies three trust-control dimensions: trustworthiness based on technical performance and human credibility; ambivalence toward autonomy, acknowledging benefits yet preferring human oversight in uncertain scenarios; and human-machine coagency, where perceived personal agency enables delegation even without intervention capability.
17 - Prediction and analysis of radiated noise from a small underwater vehicle
Authors: Fan Xu, Yun Zhao, Yu Yao, Xiangyu You, Hefeng Zhou Small underwater vehicles, valued for their stealth, compact size, and maneuverability in missions like reconnaissance, face a critical challenge: vibrations from internal components (e.g., motors, propulsion) transmit through the structure, radiating noise into water. This compromises acoustic stealth and heightens passive sonar detection risk. Using a frogman carrier as a case study, this research employs a coupled acoustic-structure interaction finite element model to simulate its vibroacoustic behavior. By analyzing structural vibrations and sound fields, the study identifies noise frequency spectra, directivity patterns, and dominant sources.
18 - Numerical simulation of hydrodynamic performance of AUV approaching an Underwater Charging Platform
Authors: Baowen Zhang, Yazhou Zhu, Yongquan Wang, Xiang Hu This study uses Computational Fluid Dynamics (CFD) to analyze hydrodynamic effects on Autonomous Underwater Vehicles (AUVs) during their approach and docking with underwater charging platforms. Numerical simulations characterize the flow field near the platform and assess hydrodynamic responses influenced by factors such as AUV speed, distance to the platform, and angle of attack. The findings provide theoretical guidance for ensuring the stability, safety, and attitude control of AUV motion during docking, offering valuable insights for the reliable design of related systems.
19 - Analysis of crack formation mechanism in the wedge ring area of underwater high-speed vessels' shell
Authors: Tao Kong, Wenfeng He, Lei Wang, Jian Ma Wedge ring connections are widely used in high-speed underwater vehicles but suffer from stress issues under extreme conditions like high-altitude transportation and vibration fatigue, leading to potential cracking and safety hazards. This paper employs non-linear finite element simulation to analyze the relationship between assembly and operational conditions of wedge rings and localized shell cracking. The findings align with real-world crack conditions. The study proposes improved design methods and assembly requirements for wedge ring connections, significantly reducing stress concentration and preventing crack generation and propagation under extreme conditions.
20 - WaveCodNet: A Frequency-Aware Framework for Underwater Camouflaged Object Detection
Authors: Jiajun Lin, Yuanbo Chen, Yi Le, Shangjing Sun, Hongyang Bai, Shuai Guo, Tianyu Deng Detecting camouflaged objects in Unmanned Surface Vehicles (USVs) is challenging. WaveCODNet, a novel camouflaged object detection framework, addresses this by combining frequency-domain analysis with multi-scale feature learning. The system uses an enhanced Pyramid Vision Transformer backbone and introduces three key components: a Frequency-enhanced Dense Interaction Decoder that mimics human visual perception by fusing high and low-frequency information, a Spatial-Frequency Attention Block (SFAB) that models spatial and frequency dependencies, and a Frequency-Aware Multi-Scale Fusion module that integrates multi-level features across different object scales.
21 - A Differential Time-of-Arrival Localization Method for Underwater Moving Platforms Based on Time Reference Synchronization and Motion Compensation
Authors: Xiaoyue Shao, Chuanqi Zhu, Liang An Accurate underwater localization is hindered by clock offsets and platform motion, degrading traditional TOA methods. High-cost atomic clocks mitigate synchronization errors but fail to address motion-induced propagation delays. This paper introduces a lightweight differential TOA (dTOA) framework using time-reference synchronization and motion compensation. Fixed beacons sequentially broadcast signals, creating a unified time reference without external clocks. A motion- compensation model corrects displacement-induced timing bias for moving receivers, while a novel dTOA estimator cancels residual clock biases.
24 - Parametric optimization of pore characteristics for enhanced acoustic attenuation in coated underwater shells
Author: Xiangxi Bu This study analyzes underwater vehicle acoustic coatings using a 2D vibroacoustic finite element framework with multiphysics coupling at fluid-solid interfaces. It quantifies how pore geometries (shape, scale, and spatial configuration) affect radiated sound pressure levels (SPL). Results show coatings achieve up to 50 dB attenuation between 100–500 Hz, with circular pores outperforming square and triangular pores by 50 dB above 100 Hz. Larger pore radii and multi-row designs further enhance attenuation across 0–1500 Hz.
25 - A simulation analysis of the lubrication characteristics of the planetary cylindrical roller bearing in power coupling mechanism
Authors: Xinyi Li, Ruyi Zhou, Hengtai Ni, Jing Liu This study establishes a CFD model of a cylindrical roller bearing under planetary gear train conditions, using MRF and VOF methods to analyze oil-air two-phase flow. Results show that centrifugal forces from gear revolution significantly redistribute velocity, pressure, and lubricating oil, with higher oil concentration on the outer side of the cavity and lower on the inner side. This effect intensifies with higher rotational speeds and temperatures due to increased centrifugal force and reduced oil viscosity, offering insights for optimizing lubrication system design in planetary bearings.