← (Macro-trends)Perception & sensor fusion
Every item tagged with this theme, most relevant first.
↓ (260 items)
This paper introduces a BIT*-based path-planning method for Maritime Autonomous Surface Ships (MASS) to handle dynamic obstacle uncertainties in congested waters. The approach combines collision avoidance and path replanning into a single optimization problem, using probabilistic collision risk in the BIT* cost function to prioritize safer routes. It also integrates COLREGs rules to ensure compliance with maritime navigation regulations. Simulations show the method produces safe and reliable paths despite positional uncertainty from sensor errors and communication delays.
This study introduces an autonomous ship navigation method using the OODA (Observe–Orient–Decide–Act) loop theory to improve safety and decision-making in complex maritime scenarios. The approach includes dynamic ship domain modeling, risk assessment, and collision avoidance strategies based on a three-degree-of-freedom motion model and an enhanced velocity obstacle algorithm. Simulations on OpenCPN confirmed the method’s effectiveness, maintaining safe distances and compliance with COLREGs (Collision Regulations) even in high-risk encounters. The framework provides a closed-loop system for real-world autonomous navigation applications.
This study proposes a machine learning framework using XGBoost to assess maritime collision risks for autonomous vessels, replacing the computationally intensive Dempster-Shafer model. Using AIS data, the model predicts Collision Risk Indicators (CRI) with high accuracy (95.36% R² in training, 90.21% in testing) and integrates them into collision avoidance algorithms. The approach processes high-traffic scenarios faster than traditional methods and reduces collision risk by 33.0% in autonomous-autonomous encounters and 28.7% in autonomous-conventional encounters. Results highlight improved efficiency and scalability for real-time maritime navigation safety.
This paper presents a modernized three-level hierarchical control system for Maritime Autonomous Surface Ships (MASS) to handle external disturbances like waves, wind, and currents. It proposes a modified proportional-derivative control law with corrections for roll, trim, and feedforward compensation, improving course accuracy and reliability in harsh conditions. The system integrates sensor data, processing, and control levels while monitoring seakeeping parameters without extra sensors. Its scalable and open structure allows integration with vessel traffic and motion control systems, enhancing navigation safety.
05Defence Industry Europe2026-10-05
Saab UK and Kraken have integrated the Giraffe 1X radar onto an autonomous vessel to detect drone threats early. The system aims to enhance security for ships and ports by providing advanced warning. The collaboration focuses on improving maritime surveillance capabilities. No specific operational results or testing details were provided.
The paper critiques the assumption that Maritime Autonomous Surface Ships (MASS) and port digitalization alone solve maritime transport challenges, arguing risks arise from systemic factors like vessel size, infrastructure limits, and organizational complexity. It emphasizes MASS as part of a broader socio-technical system, requiring smart ports, traffic management, and maritime simulators for safe coexistence with manned vessels. The role of simulators as cyber-physical test beds for validating operations, decision-support systems, and human-automation interaction is highlighted. The authors stress that successful MASS integration depends on managing complexity through predictive coordination, simulation-based validation, and regulatory adaptations, rather than just increasing autonomy.
07MarketScreener2026-10-05
Saab UK and Kraken developed a prototype autonomous surface vessel to counter drone threats at sea. The system integrates Saab’s Giraffe 1X radar with Kraken’s unmanned K3 Scout vessel. It is designed to detect and monitor drones in maritime environments. The collaboration aims to enhance maritime security through autonomous technology.
This paper introduces a Formal Safety Assessment (FSA) framework to evaluate risks in Maritime Autonomous Surface Ship (MASS) operations, addressing gaps in traditional methods that rely on crewed-ship data. It establishes a hierarchical risk indicator system across five dimensions—ship machinery, human factors, environment, information technology, and management—using frequency and severity indices to classify risks. A case study with 50 experts found MASS operations fall within the "ALARP" (tolerable) risk region, with environmental, IT, and machinery factors posing the highest risks. Key contributors to overall risk include environmental perception, navigation decision-making, and communication capability.
The paper introduces a High-Frequency Composite Safety Index (HF-CSI) for real-time risk assessment of Maritime Autonomous Surface Ships (MASSs), combining weather, collision, machinery health, and grounding risks into a single 0–10 scale. Using full-scale voyage data from a container ship over 6.5 days, the HF-CSI averaged 8.01, with dominant risks varying by sea area and no single hazard accounting for over 65% of risk-bearing samples. A non-compensatory weighted-maximum rule prevents critical hazards from being masked, and the method accurately flagged high-risk segments that compensatory averaging missed. The framework also differentiates collision avoidance behaviors and adjusts machinery health baselines based on propulsion modes.
10Naval Technology2026-10-06
Saab and Kraken have partnered to develop an autonomous vessel designed for drone detection. The project aims to enhance maritime security by leveraging unmanned technology. No specific technical details or operational capabilities were disclosed in the announcement. The collaboration was reported by Naval Technology on October 6, 2026.
This paper reviews research on autonomous ship and harbor maneuvering from 2015 to 2026, covering control systems, AI, sensor fusion, digital twins, and logistics. It introduces a seven-layer cyber-physical taxonomy and a maturity model to assess integration levels, from isolated vessel autonomy to fully federated systems. While progress has been made in individual areas like control and perception, gaps remain in cross-layer coordination, cybersecurity, and regulatory compliance. The authors propose a Federated Maritime Cyber-Physical Architecture to improve interoperability and resilience in autonomous harbor operations.
This paper explores cyber-attack detection for Maritime Autonomous Surface Ships using Python-based predictive analysis. It analyzes dynamic data like geo-location, timestamps, and navigation status with machine learning and deep learning techniques to identify anomalies in real-time and Automatic Identification System datasets. The proposed hybrid model combines LSTM and random forests to improve detection accuracy, reliability, and scalability. The system aims to support proactive maritime monitoring and risk mitigation for autonomous operations.
13Army Recognition2026-10-05
Kraken and Saab UK integrated the K3 SCOUT uncrewed vessel with Saab’s Giraffe 1X radar to enhance drone detection at sea. The collaboration aims to improve maritime surveillance by combining autonomous vessel technology with advanced radar systems. The Giraffe 1X radar provides real-time tracking of small, fast-moving aerial targets. This pairing is designed to bolster security in naval and coastal operations.
14Army Recognition2026-11-05
Leonardo and Shield AI signed an agreement on October 29, 2026, to integrate Shield AI’s Hivemind autonomy software into Leonardo’s Proteus uncrewed helicopter for the UK Royal Navy. The Proteus, a three-tonne aircraft developed under a £60 million UK Ministry of Defence programme, completed its maiden flight in January 2026 and is designed for anti-submarine warfare, ISR, and contested logistics missions. The Hivemind SDK will enable customized AI decision-making for these operations while Leonardo retains intellectual property rights. No integration timeline, test dates, or production orders have been disclosed.
15Hellenic Shipping News2026-10-15Cyprus Shipping News
SEA.AI's maritime machine vision technology has been integrated into MARTAC’s unmanned surface vessels (USVs), enhancing autonomous visual intelligence for missions like surveillance, security, and patrol. The system uses AI to process electro-optical and thermal data in real time, classifying vessels and objects without human intervention, addressing gaps in radar and AIS limitations. MARTAC’s USVs, including the T38 Devil Ray, demonstrated this capability during an 8-day autonomous mission in challenging conditions. The partnership aims to improve situational awareness for unmanned platforms operating beyond visual range.
This study examines Myanmar’s legal and sustainability readiness for Maritime Autonomous Surface Ships (MASS). Survey data from 51 seafarer officers shows 40% find Myanmar’s current legal framework insufficient for MASS adoption. About 70% of respondents believe MASS will improve environmental sustainability. The research aims to help Myanmar develop a systematic framework for future MASS operations in its coastal waters.
This study introduces a framework for improving the safety of maritime autonomous surface ships (MASS) by analyzing perception uncertainty and optimizing sensor layouts. It models sensor detection probabilities, state-estimation uncertainties, and collision risks using an ellipse-based safety envelope. The optimized sensor configuration achieved 68.49% mean coverage, outperforming a fixed heuristic layout (67.89%), with daytime and nighttime coverage of 74.45% and 62.56%, respectively. A synthetic scenario with ten ships demonstrated the framework’s effectiveness in evaluating sensor performance under SOTIF constraints.
This paper explores how autonomous maritime systems depend on diverse sensors like radar, LiDAR, GNSS, and INS for navigation and surveillance. Challenges such as radar clutter, jamming, and GNSS signal interference threaten reliability, requiring techniques like clutter suppression and adaptive countermeasures. Multi-sensor fusion is highlighted as a solution to reduce uncertainty and improve situational awareness by combining data from complementary sensors. The approach ensures robust decision-making even in adverse or degraded conditions.
This paper presents a multi-sensor fusion system combining RGB camera and LiDAR data to improve obstacle detection and situational awareness for autonomous ships. Camera data is processed using a pre-trained object detection model, while LiDAR data undergoes clustering and classification via a Random Forest algorithm. The system tracks dynamic obstacles using a Global Nearest Neighbour algorithm with Kalman filtering, reducing false positives from noisy measurements. Real-world tests demonstrated 92% track number accuracy in complex maritime scenarios.
This paper introduces a decentralized motion planner for autonomous surface vessels in urban waterways, using a sampling-based model predictive control method. The planner simultaneously calculates the vessel's trajectory and predicts interactions with other vessels while adhering to traffic rules through discontinuous cost functions. Tested in real Amsterdam canal maps and outdoor experiments with scaled prototypes, the method achieved collision-free and deadlock-free navigation in 780 randomized trials, outperforming optimization-based and non-interaction-aware planners. Onboard execution ran at 10 Hz, successfully handling head-on and crossing scenarios.
This study introduces a framework for autonomous ships to assess collision risks in real time, accounting for sensor uncertainty from AIS and radar data. It uses an Unscented Kalman Filter to estimate target-ship states and integrates uncertainty into a probabilistic ship domain, enabling metrics like Probabilistic Collision Exposure (PCE) and Expected Collision Energy (ECE). Simulations under noisy conditions showed target-position errors averaging 3 m, with PCE reducing post-CPA residual risk significantly compared to the Collision Risk Index. The framework maintained a 10-Hz update rate across single- and multi-target scenarios.
This paper examines challenges in autonomous ship navigation, such as emergency behavior, harsh conditions, and human error in mixed traffic. It explores how Digital Twins (DT), combined with AI and big data, could improve safety and efficiency in complex marine environments. The study reviews current autonomous ship systems and traffic models to assess DT’s potential for enhancing navigation. DT may also support decarbonization and smart mobility in maritime transport.
23Smart Maritime Network2026-11-06
Ubotica tested an AI-powered system on its CogniSAT-6 satellite that autonomously identifies vessels in imagery and schedules follow-up observations without ground station input. The satellite detected a cruise ship in the Pacific, tracked its movement over two days, and confirmed its departure—all independently. This "Autonomous Scheduling" technology could enhance maritime surveillance, particularly for vessels without AIS signals or potential threats to offshore infrastructure. It may also enable coordinated satellite operations, where one satellite cues another for follow-up observations.
This paper examines safety challenges in maritime autonomous surface systems, highlighting risks from degraded communication, unclear control authority, and delayed human intervention. It analyzes 80 documented hazard scenarios from regulations, trials, and incident reports to identify key risk areas like remote supervision, control links, and navigation interactions. The study proposes a safety engineering approach to define control constraints and recovery requirements, emphasizing the need for explicit degraded modes, fallback behaviors, and emergency coordination. Findings show safety depends more on system design and recovery protocols than autonomy level alone.
This paper introduces LDA-D3QN, an improved deep reinforcement learning method for autonomous navigation of unmanned surface vehicles (USVs) in complex obstacle scenarios. The method reduces redundant environmental data by combining target-related and local obstacle features, improving decision stability and training efficiency. Tested on a simulation platform, LDA-D3QN achieved a 91.4% success rate, outperforming other reinforcement learning algorithms like PPO and Dueling DQN. Preliminary real-world sensor data validation supports its feasibility for future autonomous navigation applications.
26shield.ai2026-10-28
Shield AI and Rheinmetall Barzan Advanced Technologies are collaborating to enhance autonomy for Intelligence, Surveillance, and Reconnaissance (ISR) on Qatari Unmanned Surface Vessels (USVs). The partnership aims to integrate advanced autonomous systems into these vessels. This initiative is focused on improving operational capabilities for Qatar. No specific technical details or outcomes were provided in the announcement.
27openalex2026-10-27Danielle Luciana Aurora Soares do Amaral, João Gabriel R. N. Ferreira, Rosa M. B. Cavalcanti, Marcelo Merten Cruz +10 moreDOI ↗ (opens in a new tab) This pilot study tested autonomous water sampling combined with eDNA metabarcoding to monitor biodiversity near an offshore FPSO in Brazil’s Santos Basin. Using an uncrewed vessel, researchers collected 64 samples, detecting 34 species with 12S and 28 with 16S markers, covering 65–74% of estimated species richness. Weak spatial differences were found around the FPSO, while short-term temporal variation was high, with taxon turnover driving most dissimilarity. The approach proved feasible for offshore monitoring, recommending repeated sampling to identify persistent ecological changes.
This paper presents a COLREGs-compliant trajectory planning framework for autonomous ships that unifies static and dynamic obstacle avoidance by parameterizing time in space. It converts dynamic obstacles into polygonal representations, integrating global and local planners while accounting for uncertainties in vessel speed and course. The method is implemented using an extended version of the SeaCharts Python library, which now supports S-57 files and weather data. Case studies demonstrate improved robustness against target ship motion uncertainty compared to velocity obstacle and scenario-based MPC approaches.
This study introduces a method to identify high-risk ships in mixed maritime traffic involving autonomous (MASS) and conventional vessels. It uses a 4D-OPTICS approach to filter ships in dense encounters and an Axiomatic Fuzzy Set (AFS) model to convert risk indicators into interpretable semantic categories like spatial proximity and motion uncertainty. A hierarchical model then prioritizes ships based on COLREGs rules and semantic risk levels. Simulations and real-world AIS data from Yingkou Port validate the framework’s effectiveness.
This paper details a real-time obstacle avoidance system for autonomous vessels, tested in a 2D maritime simulation. The system uses a proximity detection algorithm built with Pygame to adjust ship trajectories and prevent collisions with static or moving obstacles. It incorporates a ship motion model and particle system for realistic environmental representation. The approach aims to support autonomous navigation, telemetry control, and intelligent maritime transport solutions.
This study introduces a multi-platform autonomous survey framework for documenting Great Lakes shipwrecks, tested in Lake Michigan and Lake Ontario (2021–2022). The approach combines AUVs, ASVs, sonar, magnetometry, and optical imaging in a structured workflow to cover large areas efficiently. Surveys generated 19.72 km² of geophysical data, including sonar mosaics and magnetic maps, leading to the discovery of a previously unknown wooden shipwreck in Lake Ontario. The method improves efficiency, adaptability, and scalability for preserving submerged cultural heritage.
This study presents a collision avoidance framework for autonomous ships that accounts for positional uncertainty in AIS data. An unscented Kalman filter estimates the motion and uncertainty of nearby vessels in real time, improving trajectory predictions. Collision risk is assessed using a quantitative index, and avoidance maneuvers are generated via the velocity obstacle method. Simulations demonstrate a 96% reduction in estimation error, achieving ~3 m positional accuracy compared to nonpredictive approaches.
This paper introduces an adaptive mission planning method for solar-powered autonomous surface vessels (ASVs) to optimize scientific data collection. The approach uses a clarity-driven ergodic search, where clarity is a mathematically defined measure tied to information entropy, to prioritize high-interest areas identified during the mission. The system continuously updates its path based on real-time measurements and energy constraints. Simulations and field experiments at Jordan Lake, North Carolina, validate the method’s effectiveness.
34openalex2026-05-26Josep Marin Garces, Andrea Pino Jarque, Salvador López Barajas, Guillem Fustier Castillo +6 moreDOI ↗ (opens in a new tab) This study demonstrates a field test at Spain’s Port of Castellón, where a catamaran surface robot with an integrated Launch and Recovery System (LARS) deployed and retrieved a BlueROV underwater vehicle. The trial covered navigation, remote supervision, underwater operations, and recovery using an immersive augmented reality teleoperation interface. Conducted by CIRTESU and IRSLab, the experiment highlights advancements in cooperative surface–underwater robotics and autonomous maritime technologies. The work supports research on robotic inspection and intervention in port and marine environments.
This review examines visual perception for autonomous surface vehicles in challenging waterway environments, highlighting dynamic disturbances, poor lighting, bad weather, and small targets as key obstacles. It organizes real-world waterway datasets by diversity and complexity while evaluating perception methods for robustness, dataset fit, and deployment limits. The paper analyzes how environmental factors cause perception failures and links model design, data quality, and practical constraints. Emerging solutions like multimodal perception and domain generalization are proposed to improve reliability in maritime applications.
This paper examines the role of Remote Operations Centers (ROCs) in monitoring and managing Maritime Autonomous Surface Ships (MASS), particularly Level 2 and 3 vessels, to improve navigational safety in high-traffic waterways. Current ROCs depend on tools like AIS, radar, and cameras but face challenges such as data latency and radar limitations in complex areas. The authors propose a real-time monitoring system using navigation buoys equipped with cameras, AIS receivers, VHF transceivers, and edge computing to fuse and analyze data locally. The framework aims to enhance monitoring and service quality by enabling collaboration between buoys and ROCs.
This paper explores challenges in Maritime Autonomous Surface Ships (MASS), including handling large sensor-generated data onboard and transferring it remotely. It highlights the reliance on AI for decision-making and the need for localized tuning of these systems. The authors propose a transfer learning-based computing architecture to improve autonomous maritime operations. A case study demonstrates its application in monitoring autonomous vessel performance.
38portnews.ru2026-10-22
South Korea’s South Gyeongsang Province has requested approval to designate waters off Geoje Island as an autonomous-ship operating zone, enabling Samsung Heavy Industries to continue sea trials after its current regulatory exemption ends in 2026. The proposed area covers coastal waters near Hansan, Gajo, and Chilcheon islands, with a decision expected by year-end. Samsung Heavy is testing a 12-meter autonomous vessel featuring AI-based navigation, collision avoidance, and sensor systems. The province aims to support the technology while minimizing disruptions to fishing, shipping, and leisure activities.
39Smart Maritime Network2026-10-13
SEA.AI’s machine vision technology has been integrated into MARTAC’s unmanned surface vessels (USVs), enabling real-time AI-based detection and classification of surface objects using electro-optical and thermal sensors. The system autonomously identifies vessel types, headings, and proximity without operator input, enhancing USV capabilities for missions like the T38 Devil Ray’s eight-day autonomous operation in Sea State 5 conditions. MARTAC and SEA.AI emphasize the technology’s role in closing gaps for fully autonomous maritime operations. The integration is part of MARTAC’s focus on advanced sensor solutions for its USV range, including the MANTAS and Devil Ray models.
This paper introduces the Segmented Autoregressive Global Attention Mechanism Transformer (SAGM-TF) to improve multi-step vessel trajectory prediction for maritime collision avoidance. The framework combines an optimized kinematic encoding module, segmented autoregressive decoding, and a cross-segment global attention mechanism to reduce error accumulation over a 90-second horizon. A new trajectory dataset, incorporating heading angle and angular velocity, was created to better capture vessel maneuvering behavior. Experimental results show the method outperforms traditional approaches in accuracy and robustness for complex maritime scenarios.
The paper introduces a modified particle filter for maritime target tracking using autonomous surface vessels (USVs) in scenarios with intermittent communication. The method adjusts particle distribution to prevent overconfidence in local estimates when external data is unavailable. Tests with USVs on a lake show the approach maintains performance under normal conditions while improving results in certain edge cases. The work focuses on distributed consensus without centralized coordination.
This paper introduces a LiDAR-based deep reinforcement learning (DRL) framework for autonomous surface vessels (ASVs) to follow paths and avoid static obstacles in narrow waterways. Using a Soft Actor-Critic (SAC) policy trained in simulation, the system processes LiDAR data and motion states to control rudder and propulsion, achieving a 95% success rate in 500 simulated trials with low collision rates. Field tests with a model-scale vessel demonstrated obstacle avoidance but showed increased path oscillations and cross-track errors compared to simulation. The approach relies on compact LiDAR sector features and continuous control commands for real-time navigation.
This paper reviews key sensor technologies used in Maritime Autonomous Surface Systems, explaining their functions and integration within autonomous navigation. It emphasizes the importance of sensor data fusion to improve environmental awareness. The authors discuss recent advancements and future trends, particularly quantum inertial and magnetic sensors, which are expected to transform navigation technology. The focus is on their potential impact in the coming years.
44shipmanagementinternational.com2026-04-02
Orca AI has been granted ClassNK Technology Qualification for its autonomous ship perception capabilities. This qualification recognizes Orca AI's technology for enhancing safety in autonomous maritime navigation. The approval highlights the system's ability to detect and avoid obstacles effectively.
45hromadske2026-11-04Hromadske
A German defense startup, Stark Defence, plans to launch the unmanned vessel *Vanta 24* in 2027 to monitor Russia’s shadow fleet and protect undersea infrastructure in the Baltic and North Seas. The 24-meter vessel can carry up to 50 metric tons of payload, including weapons, and operate at speeds over 64 km/h without a crew. Stark claims the vessel will be far cheaper and faster to produce than traditional warships, with deployment possible within months. The project aims to counter Russian hybrid threats, though analysts note challenges in scaling up the technology.
46defence-industry2026-10-25
EDGE demonstrated integrated autonomous systems for the UAE and U.S. navies, combining unmanned surface vessels, AI-driven command-and-control, and counter-drone defenses under a unified command structure. The exercise used TACTICA AI’s Naval Warfare C2 Centre, HADAF targeting systems, and assets like the M-DETECTOR vessel and VORTEX-E interceptor to enhance interoperability. GPS anti-jamming and multispectral sensors supported detection and tracking during reconnaissance and force protection missions. The collaboration aimed to improve coordination between allied forces through shared autonomy and operational experience.
47영남신문2026-10-22
A 2026 event showcased future maritime technologies, including AI-powered autonomous ships and underwater robots. The exhibition took place inland, highlighting advancements in unmanned and robotic systems for marine applications. These innovations aim to improve efficiency and safety in maritime operations. The event was reported by *Yeongnam Ilbo*.
48헤럴드경제2026-10-21
The Korea Ship and Ocean Engineering Research Institute presented advanced maritime technologies, including AI-powered autonomous vessels and underwater robots. The showcase highlighted future developments in ocean engineering. No specific technical details or outcomes were disclosed. The announcement was reported by 헤럴드경제 on October 21, 2026.
49세종경제뉴스2026-10-21
A 2026 event showcased future maritime technologies, including AI-powered autonomous ships and underwater robots. The exhibition took place inland, highlighting advancements in unmanned and robotic systems for marine applications. These innovations aim to improve efficiency and safety in maritime operations. The event demonstrated emerging trends in the industry.
50Zone Militaire2026-10-09
Exail, a French company, introduced a new maritime surveillance system using its Oryx naval drone range. The system aims to enhance monitoring capabilities at sea. Details on specific features or operational use were not provided. The announcement was made on October 9, 2026.
This study introduces SGNav, a framework enabling Autonomous Surface Vehicles (ASVs) to navigate toward targets described in natural language rather than predefined coordinates. The system combines semantic grounding, harbour-specific filtering, and CLIP-based verification to interpret descriptions, filter distractions, and execute control policies. Experiments in simulated port environments showed success rates of 90–97% across tasks, with high semantic accuracy and low error rates. Ablation studies confirmed the critical role of semantic grounding and filtering for reliable performance.
52Smart Maritime Network2026-11-06
Jiangsu Ocean Shipping is testing autonomous technology on two 120 TEU electric container ships, *Jiangyuan Jinling* and *Jiangyuan Qinhuai*, along an 80 km stretch of the North Jiangsu Canal. The vessels will use radar, AIS, and high-definition sensors to navigate varying conditions, including bends and bridges, with tailored navigation settings for each environment. The project aims for 90% object recognition accuracy and expects autonomous navigation to cover at least one-third of the trial distance. A digital twin and shore-based interface will assist with voyage planning, risk alerts, and port coordination.
The paper introduces QSD-STFormer, a quaternion ship domain-guided spatiotemporal transformer for ship trajectory prediction in dense maritime environments. It combines a CVAE framework with a QSD-guided Interaction Prior Module to incorporate physical navigation constraints, improving data-driven modeling. A Spatiotemporal Joint Modeling module captures dynamic ship interactions, while a GMM-Latent Representation module enhances prediction diversity. Tests on AIS data from the Port of New York and Wuhan Riverfront show a 15.28% reduction in minADE and 24.15% in minFDE compared to existing methods.
54crossref2026-09-08Sergey I. Kondratyev, Evgeniy V. Khekert, Nikita V. Martyushev, Boris V. Malozyomov +5 moreDOI ↗ (opens in a new tab) A risk-aware multisensor fusion system was developed for autonomous navigation and mooring in confined waters, integrating data from RTK-GNSS, inertial sensors, radar, LiDAR, cameras, and other sources. The system uses fault detection, sensor-quality assessment, and dynamic authority allocation between automation and operators to maintain reliability. Testing over five days with 40 runs and 480 evaluation epochs showed horizontal-position and heading errors of 0.043 m and 0.176°, respectively, under normal conditions. The system detected all 64 fault events, with 99.3% protection-level coverage for single faults and 100% for combined faults, ensuring safe operation even under degraded sensor conditions.
This study introduces SLC-Fusion, a multimodal perception system for Maritime Autonomous Surface Ships (MASS) combining shipborne LiDAR, cameras, IMU, and RTK data. It improves sensor calibration, motion compensation, and LiDAR denoising while using AI (YOLO 11s-seg) and spatial clustering to detect and localize port targets in 3D. Real-world tests show high accuracy with a minimum ranging error of 0.07 m, a mean IoU of 67.0%, and fast processing times (5.9 ms for images, 4.6 ms for point clouds). The method also establishes a port navigation perception dataset.
This paper presents a stereo-vision system for ship detection and localization on Unmanned Surface Vehicles (USVs). The system uses GS-YOLO, an improved version of YOLO11n with added modules (GLSA, BiFPN, and SIoU loss), to enhance small ship detection accuracy while reducing model parameters by 17.8%. For localization, a cascaded filtering algorithm combining median and Kalman filtering processes depth data to minimize errors caused by dynamic interference. Tests show GS-YOLO achieves 93.5% mean average precision, and the localization method reduces positioning error by up to 89.3% within 50 meters.
This study introduces a hybrid deep learning model combining CNNs, Transformers, and MLPs to improve ship trajectory prediction using AIS data. The CNN extracts local motion patterns, while the Transformer captures long-range temporal dependencies, and the MLP decodes features for accurate forecasts. K-means clustering enhances model generalization by adding static navigational features. The model outperformed baseline methods, achieving a root mean square error of 1.56 km and a mean absolute error of 1.07 km at a 60-minute prediction horizon.
58112.ua2026-11-04
The Vanta 24 is an unmanned vessel developed by Stark to protect undersea cables in European waters. The vessel is designed to monitor and safeguard critical infrastructure from potential threats. Its deployment aims to enhance security for communication and energy networks. No specific operational details or outcomes are provided in the text.
SeaTrack-DeepSORT is a vision-based framework for ship detection and multi-object tracking in complex maritime environments, addressing challenges like scale variations, occlusions, and camera motion. It combines SeaDet-YOLO, an enhanced YOLOv11 detector with ADown downsampling and a P2 detection head, with a modified DeepSORT tracker that uses confidence-dependent updates and motion-compensated association. The system improves detection and tracking accuracy while ensuring stable trajectories. Experiments on maritime datasets demonstrate its effectiveness for autonomous perception in unmanned surface vehicles.
This study explores factors influencing Indian tourists' acceptance of Autonomous Maritime Tourism (AMT), an AI-driven innovation in smart ports and navigation. Using the Technology Acceptance Model and Diffusion of Innovation theory, it surveyed 237 Indian tourists and analyzed data via discriminant function analysis. Social influence emerged as the strongest predictor of behavioral intention, followed by relative advantage, perceived complexity, and facilitating conditions. The findings highlight that social validation and perceived benefits outweigh risk factors in shaping AMT adoption in India.
61The Defense Post2026-09-14
Moog successfully demonstrated its Reconfigurable Integrated-weapons Platform (RIwP) turret on Voltaic Marine’s autonomous vessel. The test showcased the turret’s compatibility with unmanned maritime systems. The RIwP turret is designed for modular weapon integration and adaptability. The demonstration took place in 2026 as part of defense technology advancements.
This paper addresses challenges in maritime search and rescue (SAR) by developing vision-based human detection systems for Unmanned Surface Vessels (USVs). Current SAR operations face delays due to reliance on manned crews, while USVs struggle with limited water-level vision, navigation, and data transfer issues. The authors created new datasets featuring people in water to improve machine vision training and integrated a vision system into a fully autonomous USV. The work aims to enhance the effectiveness of SAR USVs by overcoming existing technological limitations.
The paper discusses the MONUSEN project, an EU-funded initiative enhancing public engagement in marine robotics to support UN Sustainable Development Goal 14 (Life below Water). It involved two public activities: demonstrating MARUS, a simulator for realistic maritime environments, and allowing participants to pilot the SWAMP autonomous surface vehicle (ASV) for data collection. These activities successfully engaged diverse groups, improving public understanding of autonomous vehicle control and marine technology. The project bridged gaps between citizens and scientists, fostering interdisciplinary collaboration.
This paper introduces an Autonomous Guided Vehicle (AGV) that uses Vision-Language Models to navigate indoor spaces based on natural language commands. The system, built on a Raspberry Pi 5 with a camera, ultrasonic sensors, and motors, interprets spoken instructions and combines visual and language data to identify targets and avoid obstacles. Tests in various indoor settings showed high accuracy in command recognition and reliable navigation performance. The approach improves flexibility over traditional fixed-route AGVs in dynamic environments.
This paper introduces ODDIT, a digital twin-based method for detecting out-of-distribution (OOD) states in autonomous vessels (AVs) before they occur. The approach uses two machine-learning models to predict future vessel states and identify potential OOD conditions, enabling proactive interventions. Evaluated on five vessels under simulated conditions with noise and environmental disturbances, ODDIT achieved 99% accuracy in OOD detection, measured by AUROC and TNR@TPR95 scores. The method supports predictive maintenance, fault diagnosis, and scenario-based testing for AVs.
66Marine Insight2026-05-25
Germany has launched MODULARIS, a new floating laboratory designed to test autonomous ship technologies and advanced propulsion systems. The facility will serve as a research platform for future maritime innovations. The project aims to accelerate the development of smart and sustainable shipping solutions. No specific testing results or methods have been disclosed yet.
The paper *Sim2Sea* introduces a framework to improve autonomous maritime navigation in congested waters by addressing the gap between simulation and real-world performance. It features a GPU-accelerated simulator, a dual-stream spatiotemporal policy with safety mechanisms, and domain randomization to enhance real-world transfer. Tests show the method achieves faster training and safer navigation than existing approaches. The policy, trained only in simulation, successfully operated a 17-ton unmanned vessel in real congested waters without additional adjustments.
68CPG Click Petróleo e Gás2026-05-22
The Chinese autonomous ship V180 is 24 meters long and operates without a crew in open waters. It can independently launch and recover ROVs, with a depth capability of up to 3,000 meters. The vessel can perform missions for up to 30 days without human intervention.
69Digital Ship2026-04-30
NYK successfully demonstrated the real-world integration of autonomous vessel systems in Singapore. The trial showcased the practical application of these systems in a busy maritime environment. The initiative highlights advancements in autonomous shipping technology. The project aims to improve safety and efficiency in maritime operations.
70搜狐网2026-11-04
The Ouka Smart Ship received China's first type approval certificate from the China Classification Society (CCS) for its "Unmanned Boat Perception and Autonomous Navigation System." This certification marks a milestone in China's maritime industry for unmanned vessel technology. The system enables autonomous navigation and environmental perception for unmanned boats. The achievement highlights progress in smart shipping development.
71파이낸셜뉴스2026-11-02연합뉴스
Korea Maritime and Ocean University is leading the development of standardization for autonomous vessel situational awareness systems. The initiative aims to establish consistent guidelines for these systems. The project is expected to enhance safety and efficiency in autonomous maritime operations. No specific technical details or outcomes were provided in the report.
72한국대학신문2026-11-02
Korea Maritime and Ocean University is leading the development of international standards for camera-based situational awareness in autonomous vessels. The university is spearheading efforts to establish global guidelines for this technology. This initiative aims to improve safety and navigation for unmanned ships. The work is expected to influence future maritime regulations.
73부산일보2026-11-02
Korea Maritime and Ocean University is leading the development of international standards for camera-based situational awareness systems in autonomous vessels. The initiative aims to establish unified guidelines for these systems. This effort positions the university as a key contributor to global maritime technology standards. The project reflects advancements in autonomous shipping safety and regulation.
PAC-Net is a new framework for maritime image dehazing, addressing challenges like uneven haze distribution, blue-green color casts, and cross-domain variability in marine environments. It introduces an Adaptive HazeData (AHD) generation method for realistic haze synthesis, a Color-Aware Channel Attention (CCA) module to correct color imbalances, and a progressive learning strategy to handle diverse scenarios. Tested on the Adaptive Maritime Haze Dataset (AMHD) and RESIDE benchmark, PAC-Net outperforms existing methods in structural and color restoration. The approach demonstrates strong cross-domain generalization.
75매일신문2026-10-28
A South Korean initiative aims to develop fully unmanned autonomous vessels by 2032. The project will integrate AI into ports and fishing grounds to enhance operations. The focus includes improving navigation, safety, and efficiency in maritime activities. The plan reflects broader efforts to modernize the shipping and fishing industries.
76헤럴드경제2026-10-28
South Korea aims to become the world leader in autonomous ships by 2035, with crewless vessels navigating the seas and AI managing container operations. The country is advancing technology to enable unmanned maritime transport. Challenges remain in achieving full autonomy and global regulatory alignment. The initiative reflects efforts to modernize shipping and enhance competitiveness.
77gCaptain2026-10-21
The U.S. Navy is advancing unmanned surface vessels (USVs) from testing to operational use, with recent deployments in the Persian Gulf and alongside the USS Theodore Roosevelt Carrier Strike Group. Successful missions include rescuing downed Army aviators and supporting mine clearance, demonstrating USVs' role in real-world operations. Senior leaders, including the Secretary of War and Navy officials, are prioritizing integration into fleet operations, emphasizing organizational and training reforms. The Navy must now develop clear concepts of operations (CONOPS) to define missions, ownership, and sustainment while continuing experimentation without treating it as a standalone strategy.
78Splash2472026-10-19
A British autonomous vessel successfully tested navigation without relying on GPS. The trial aimed to demonstrate alternative positioning and guidance systems for maritime operations. The experiment highlights advancements in autonomous shipping technology. No further technical details or outcomes were specified in the report.
This paper reviews technologies enabling agricultural robots to autonomously follow crop rows for tasks like weeding and harvesting. It examines perception methods (vision, LiDAR, and multimodal fusion) and their effectiveness under challenging conditions like lighting changes and weed interference. The review also covers navigation techniques, state estimation, path planning, and motion control, comparing their accuracy, real-time performance, and adaptability. It provides a framework for integrating perception, planning, and long-term operation in complex agricultural settings.
80Marine Link2026-10-14
Voyant demonstrated GPS-denied navigation technology on an autonomous vessel for the UK Ministry of Defence (MOD) in October 2026. The system enables the vessel to operate without relying on satellite-based GPS signals. This advancement supports navigation in environments where GPS may be unavailable or compromised. The trial highlights progress in autonomous maritime technology for defense applications.
This paper presents a framework using CycleGAN-turbo for one-step unpaired image translation to generate synthetic maritime data, addressing the lack of diverse training data for Maritime Autonomous Surface Ships (MASS). The method incorporates zero-convolution skip connections to preserve small navigational details, such as distant vessels, during domain translations like Day-to-Foggy or Day-to-Night. A dataset of 7,000 maritime images was used to train and evaluate models, showing strong structural retention in foggy and sunset conditions but challenges like semantic hallucination in night translations. The approach aims to overcome data scarcity in autonomous maritime navigation by efficiently synthesizing realistic atmospheric conditions.
This study examines a 2022 ship collision in Nagoya Port caused by miscommunication via VHF radio and human error, particularly misunderstandings about vessel maneuvers at low speeds. It proposes a "digital passing agreement" system using AIS to convert verbal agreements into machine-readable data, displayed on navigation systems to improve clarity and compliance. The system detects deviations from agreed paths and integrates with Vessel Traffic Services for alerts and third-party input. The approach aims to reduce human error and improve situational awareness in port operations.
This paper presents a unified, safety-constrained optimization framework for autonomous mining systems, integrating sensor fusion, edge intelligence, and digital twin-based decision support for surface and underground environments. The approach uses mathematical modeling and multi-objective optimization to improve vehicle dynamics, drilling mechanics, and fleet coordination under GNSS-denied and extreme Nordic conditions. Validation through Monte Carlo simulations and statistical testing showed productivity gains of ~24.3%, energy reductions of ~12.8%, and safety risk decreases of ~48.6% compared to baseline systems. Key performance drivers identified include localization accuracy, communication delay, and optimization weighting.
This study introduces an image-based method to recognize nighttime encounter situations for unmanned surface vessels (USVs) by analyzing vessel navigation lights. A 3D model of navigation lights and encounter scenarios was created following COLREGs, and a dataset was generated through simulations. The method extracts features from navigation-light images and achieves classification accuracies of 96.6%–98.3% in simulations. Field tests with a small USV demonstrated 96.2% accuracy for light recognition and 94.94% for encounter situation recognition.
This study introduces TCN–KAN–PINN, a framework combining a multi-scale Temporal Convolutional Network (TCN), a Linear-KAN predictor, and physics-informed training to improve vessel trajectory prediction. The model extracts temporal features from historical AIS data, corrects nonlinear maneuver deviations, and enforces kinematic consistency using a time-discretized residual and adaptive loss balancing. Tested on real-world AIS data, it outperformed existing methods in accuracy, speed, and course consistency while maintaining efficient inference. Ablation studies confirmed the benefits of kinematic supervision for short-term predictions and the KAN pathway for longer horizons.
This study introduces a framework converting water-segmentation masks into navigation boundaries, centerlines, and corridor metrics for inland autonomous surface vehicles. Tested on 173 low-resolution images, the proposed Edge U-Net outperformed other models, achieving an IoU of 0.964 and a Boundary-F1 score of 0.685. The evaluation highlights discrepancies between pixel-overlap scores and boundary accuracy, using additional indices like the Navigation Robustness Score. The framework focuses on reproducibility and geometric assessment but does not ensure obstacle-aware or closed-loop navigation safety.
87comparethecloud.net2026-10-05
Britain should prioritize autonomous technology investments in harbor operations over ocean crossings, focusing on defined transport tasks like container transfers between terminals. Singapore’s autonomous feeder vessel program serves as a model, emphasizing reliability, safety, and integration with existing workflows rather than immediate cost savings. British suppliers could contribute by addressing specific challenges such as degraded visibility, cybersecurity, or terminal integration, but must clarify responsibilities and boundaries in contracts. Regulatory approval in the UK depends on vessel size, operating area, and use, requiring early engagement with the Maritime and Coastguard Agency to establish a clear pathway.
88Smart Maritime Network2026-10-05
Seasats has deployed a fleet of Lightfish uncrewed surface vessels (USVs) in the Gulf of Mexico to enhance US maritime border security. The solar-powered USVs, measuring 3.4 meters long and weighing 154 kg, will conduct persistent autonomous surveillance to detect illegal activities like drug trafficking and unauthorized vessel entry. Equipped with modular sensors, including cameras and sonar, the vessels can operate for months alongside crewed patrols. The deployment expands on previous use along the Rio Grande since January 2026, aiming to provide continuous coverage in areas too vast for crewed assets alone.
89Smart Maritime Network2026-09-16
Robosys has integrated its autonomous navigation technology with FarSounder sonar systems. The collaboration aims to enhance maritime situational awareness and obstacle avoidance. This integration allows vessels to operate more safely in complex environments. The solution is designed for both commercial and defense applications.
This study assesses MEMS inertial sensors for autonomous naval platforms, comparing low-cost and tactical-grade sensors under calm and rough sea conditions. Low-cost sensors show rapid error growth without external aiding, while tactical-grade sensors offer better stability. An Extended Kalman Filter (EKF) reduces drift by 60–80% when combined with periodic GNSS updates. Performance declines in high sea states due to increased noise and nonlinear motion.
This study presents a machine-learning maritime security system using YOLOv8 for pirate detection in low-light and adverse weather, trained on an augmented dataset. The system assesses threat levels by analyzing vessel size, speed, and heading to differentiate pirate skiffs from benign boats, reducing false positives. It calculates optimal evasive maneuvers, finding zigzag patterns most effective for unarmed ships. The framework operates in real-time on standard hardware, offering a proof-of-concept for early threat mitigation.
This paper examines how AI and autonomous systems in marine environments redistribute responsibility between humans, algorithms, and ecosystems. It introduces an *Ethics of Autonomy* framework, combining posthuman design theory, Actor–Network Theory, and explainable AI to balance human oversight with machine decision-making. Three case studies—FloatyBoats, Seaswarm, and Vatoz—demonstrate how adaptive robotic systems shift from monitoring to active ecological intervention. The authors argue that ethical innovation requires transparent, interpretable, and accountable algorithmic infrastructures.
This paper proposes a new architecture for situational awareness in Level 4 autonomous marine vessels, addressing limitations in traditional data fusion methods that fail to differentiate sensor failures, accuracy issues, or cyberattacks. The system includes a non-switchable core of essential sensors, a Sensor Trust Manager to isolate unreliable data sources, and a navigation scenario classifier that adjusts sensor confidence weights dynamically. The approach shifts from data fusion to "source trust management," introducing meta-awareness to monitor both the environment and sensor reliability. The design aims to comply with maritime safety regulations while improving unmanned navigation safety.
This study introduces a data-efficient method for classifying Aids to Navigation (AtoN) using vision-language models (VLMs) and Region of Interest (RoI) processing for Maritime Autonomous Surface Ships (MASS). It addresses challenges like limited labeled data and background interference by comparing a YOLOv12 classifier with CLIP, which uses prompt engineering and few-shot tuning. Tests on simulated and real-world datasets show CLIP performs robustly with minimal training data, even in low-visibility conditions. The approach improves practical AtoN classification in maritime settings.
95openalex2026-10-29Shaffiyah Cantika Kamia, Suharyono Soemarwoto Hadiningrat, Jones Zenas Rante, Rudianto Rudianto +1 moreDOI ↗ (opens in a new tab) This study explores the adoption of Unmanned Lighthouse Systems in Banjarmasin, Indonesia, highlighting benefits like a projected 40% cost reduction, improved safety, and compliance with IMO and IALA standards. Challenges include environmental constraints, regulatory gaps, digital skill shortages, unreliable power and telecoms, and cybersecurity risks. Researchers used interviews, observations, and document analysis to develop a three-phase implementation roadmap. The findings offer a socio-technical framework for modernizing maritime navigation in developing nations.
96YouTube on MSN2026-10-25
The Armada project by Ocean Infinity deploys a fleet of autonomous ships for deep-sea exploration and ocean floor mapping. These advanced vessels use smart technology to navigate and collect data from uncharted depths. The initiative aims to improve understanding of the ocean while reducing human risk in hazardous environments. The project represents a significant step in modern maritime innovation.
WADNet is a weather-adaptive framework designed for marine ship instance segmentation, addressing challenges like fog and rain that degrade image quality. It uses a Weather-Adversarial Feature Decoupling (WAFD) mechanism to reduce reliance on weather-related visual cues during training without adding inference costs. A Weather-Adaptive Dynamic Routing Mechanism (WADRM) adjusts feature transformations based on weather conditions, while Inner-CIoU improves bounding-box accuracy. Tested on a mixed-weather dataset, WADNet achieved a 63.40% Mask AP, outperforming YOLO11n-seg by 3.18 percentage points.
The paper introduces *Ship3DNet-ShipNP*, a multimodal dataset designed for maritime autonomous surface ships (MASS) to improve navigation in complex port environments. The *ShipNP* dataset includes fully annotated and synchronized spatiotemporal data to evaluate the proposed *Ship3DNet* model. Port traffic complexity is divided into three levels based on ship density: simple (≤5 ships), normal (6–10 ships), and complex (>10 ships). Navigation tasks are categorized as either berthing or underway.
99Digital Ship2026-09-17
Robosys and FarSounder collaborated to enhance autonomous navigation systems with underwater perception technology. The integration aims to improve situational awareness for unmanned vessels. This advancement supports safer and more efficient maritime operations. The partnership was announced in a 2026 industry update.
100Ocean News & Technology2026-09-14
Robosys Automation has integrated its VOYAGER AI platform with FarSounder’s sonar technology to enhance autonomous navigation for vessels. The collaboration aims to improve obstacle detection and situational awareness in maritime environments. This integration combines AI-driven decision-making with advanced sonar capabilities. The system is designed to support safer and more efficient autonomous operations at sea.
This study presents an improved YOLOv8-based method for detecting and tracking occluded ships in maritime environments. The approach enhances the model with an SEConv structure using Swish activation, integrates a CBAM attention mechanism, and fuses AIS trajectory data to address occlusion and tracking interruptions. Tested on the SeaShips dataset and a custom occlusion dataset, the method achieves stable mAP gains and improves multi-target tracking accuracy (MOTA) and IDF1 by 6.3% and 8.1%, respectively. The system reduces occlusion reconstruction error to under 1.4 seconds, demonstrating improved performance in complex maritime scenarios.
102Unmanned Systems Technology2026-09-01
Cambridge Pixel will demonstrate software and sensor integration for autonomous vessels in an upcoming showcase. The technology focuses on enhancing navigation and operational capabilities for unmanned maritime systems. The event highlights advancements in autonomous vessel control and sensor data processing. No specific technical details or outcomes are provided.
103TipRanks2026-09-01
Seasats demonstrated its autonomous vessel technology at the Lakebed 2030 Conference on September 1, 2026. The event highlighted the capabilities of uncrewed surface vessels in underwater mapping and data collection. The showcase aimed to advance discussions on autonomous maritime systems. No specific technical details or outcomes were provided in the announcement.
104Dealroom2026-08-26
Cambridge Pixel will showcase its autonomous vessel sensor software at SMM 2026. The software is designed to enhance navigation and situational awareness for unmanned ships. The demonstration aims to highlight advancements in maritime autonomy technology. The event is scheduled for 2026.
This study presents a fire detection system for autonomous ships using a four-channel model combining RGB, infrared (IR), and ultraviolet (UV) sensors with LiDAR for localization. The system uses an Xception-based network for RGB+UV images and a dense-based network for UV time-series data, reducing false alarms from fire-like sources like welding. A YOLOv7 model with LiDAR projection improved fire source localization accuracy over traditional triangulation methods. Tests in open and occluded environments confirmed the system's effectiveness.
This paper introduces AIMAPP, a framework for autonomous robot navigation that combines mapping, localization, and decision-making using a single generative model inspired by animal navigation. The system builds and updates a sparse topological map in real-time, learns state transitions dynamically, and plans actions by minimizing Expected Free Energy to balance exploration and goal-directed behavior. Tested in real and simulated environments, AIMAPP operates without pre-training, adapts to sensor noise and failures, and outperforms existing planning methods. It is designed to be modular, self-supervised, and compatible with various sensors and robot configurations.
SkyPASS is a celestial navigation sensor by Polaris Sensor Technologies that enables autonomous systems to operate in GPS-denied environments. It uses star-field imaging, solar/lunar tracking, and polarization-based sensing to provide continuous, passive navigation for ground, air, and sea applications. Real-time algorithms fuse celestial data with inertial and environmental sensors to prevent drift and maintain accuracy under dynamic conditions. Testing across ground, airborne, and maritime platforms demonstrated reliable performance despite partial sky occlusion and magnetic interference.
108TipRanks2026-06-06
Saronic Technologies demonstrated an autonomous vessel designed for U.S. port security applications. The showcase highlighted its potential use in monitoring and protecting port infrastructure. The technology aims to enhance maritime security through unmanned operations. No specific technical details or deployment timelines were provided.
GeoShield is a multi-layered defense framework designed to detect and correct adversarial tampering in autonomous vehicle navigation maps. It uses OpenStreetMap metadata and NASA SRTM elevation data to identify attacks like label flipping, road-direction manipulation, and speed-limit spoofing. The system employs heuristic validation, graph-based anomaly detection (e.g., Isolation Forest, HMMs, GraphSAGE), and selective re-verification via the Overpass API. Tested on European and U.S. road networks, it aims to flag and confirm tampered map segments with high accuracy.
The study introduces a new method for modeling and managing high-density bathymetric data using the S-100 framework, specifically the IHO S-102 standard. It employs a relational database and a three-level indexing system to organize and retrieve data efficiently throughout its lifecycle. The approach includes convex hull-based geometric constraints for interpolating water depth data and an HDF5-based storage model with block storage and version control, reducing storage space by 83.6%. Field tests in the East China Sea confirmed the generated S-102 data met international standards and achieved high terrain restoration accuracy.
111Idéal Investisseur2026-11-03
Eviden, a branch of Atos, introduced SMX, a speed sensor designed for naval drones and manufactured in France. The sensor aims to enhance navigation and operational capabilities for unmanned maritime systems. The launch reflects efforts to bolster domestic technology in defense and maritime applications. No specific technical details or performance metrics were provided.
This study presents a method for autonomous endovascular navigation using 2D image feedback from DSA and X-ray fluoroscopy. The system converts vessel images into a centerline graph, tracks device-tip progress, and uses a rule-based controller for navigation tasks like delivery and branch selection. Tested on a robotic platform with 65 runs, it achieved over 95% guidewire progress in all cases, with only 3.1% failing due to excessive stops. The approach shows promise but requires further testing with clinical images and varied anatomies.
This study compared two vision-based methods for estimating dry-material inventory in autonomous labs: RGB/YOLO segmentation and RGB-D surface reconstruction using an Intel RealSense D405. Across three materials in small tubes, the D405 method achieved lower error (3.26% mean absolute error) than YOLO (7.14%) but had slightly lower valid-result coverage. Both methods performed worse with surface disturbances, though YOLO degraded more sharply. The findings suggest selecting methods based on error rates, calibration needs, and resource requirements, with limitations tied to the tested materials and setup.
This paper evaluates calibration methods for electromagnetic logs (EMLogs) used in autonomous surface vehicles when Doppler velocity logs (DVLs) are unavailable. It compares four calibration models using two estimation pipelines on 221 minutes of real-world data from the MARVEL vehicle. Results show the bias and scale error model with a Kalman filter improves speed estimation by 71%. Dynamic maneuvers, rather than straight-line paths, further enhance calibration accuracy, enabling real-time online EMLog calibration.
This paper introduces a new motion planning framework for complex articulated autonomous underwater vehicles (AUVs) to enable safe navigation and active perception in dynamic, uncertain environments. The method adapts an existing pipeline for holonomic AUVs to handle high-dimensional, articulated systems with unknown dynamics and multiple sensors. Simulations using an underwater snake robot and an underwater vehicle-manipulator system demonstrate real-time, efficient, and safe performance. The study also highlights computational challenges and potential improvements.
This paper introduces VIMF-YOLO, a deep learning model for maritime ship detection that fuses visible and infrared images to overcome single-modality limitations. The model, based on YOLO v8, uses a dual-modal fusion module (DMFM) and efficient multi-scale attention (EMA) to enhance feature extraction and representation. A paired dataset of visible and infrared ship images was created to test the model, which demonstrated higher detection accuracy compared to state-of-the-art algorithms. The approach leverages complementary strengths of both modalities for improved performance.
The paper introduces TerrainFormer, a framework combining a world model for terrain prediction with a decision transformer for off-road navigation. It uses self-supervised pretraining on LiDAR data to learn terrain features, followed by behavioral cloning for action selection. The model achieves 87.31% test accuracy and 0.7948 macro F1 on the RELLIS-3D dataset, with strong cross-dataset generalization. Predicted future frames show minimal accuracy loss (0.79%) compared to ground-truth observations.
This paper introduces an intelligent guidance system for autonomous vessels, combining traffic rule awareness and semantic reasoning to enable collision-free navigation. The framework uses qualitative traffic rule data, environmental information, and sensor inputs to make real-time path planning decisions. A finite-state machine assigns traffic roles, while a path planner calculates safe routes considering vessels and infrastructure. The approach was validated in case studies involving head-on and crossing scenarios in short-sea and inland waterways.
This study evaluates three geometric methods—bounding boxes, convex hulls, and concave hulls—to improve ship localization accuracy in radar PPI images using the DLR Baltic Sea dataset. After compensating for a 4.5° azimuth offset, concave hull centroids achieved the lowest average localization errors (58–70 m) when validated against AIS data. The method outperformed bounding boxes by up to 12.7% and convex hulls by 2.1%, better handling irregular echo shapes caused by sea clutter and noise. It provides a lightweight, real-time solution for Vessel Traffic Services and autonomous surface vehicles.
This paper introduces a framework that jointly optimizes target perception and disturbance estimation for unmanned surface vessels (USVs). It combines a hybrid CNN-Transformer model for robust visual feature extraction, a reinforcement learning-tuned sliding mode observer for adaptive disturbance estimation, and a confidence-guided strategy to synchronize perception and observer performance. Tests in simulations and outdoor lake experiments show improved accuracy, convergence, and stability over traditional decoupled methods. The study validates initial real-world performance but notes the need for broader open-water and multi-vessel testing.
Researchers at RWTH Aachen University are developing a structured safety concept for integrating autonomous machinery in surface mining, addressing the lack of standardization in mixed-traffic operations. A key component is an integrated safety control system relying on perception sensors, though their performance under harsh mining conditions remains poorly understood. To study this, a long-term test stand was built at a quartz sand open-pit mine to evaluate sensor reliability across varying environmental influences. The work aims to improve safety and efficiency in autonomous mining operations.
WorldSense is a framework designed to improve autonomous vehicle safety in rare or high-risk driving scenarios. It combines predictive world modeling, anomaly detection, and safety-aware planning by encoding the environment with a CNN-based system and predicting future scenes using a recurrent memory module. The system detects anomalies in real time and adjusts trajectories to avoid collisions. Tests on CARLA and nuScenes show a 34.2% reduction in collisions and a 14.6 percentage point increase in route completion compared to transformer-based baselines.
This paper presents an autonomous navigation system for a split-type flying vehicle's intelligent chassis, using LIDAR and IMU for positioning and the A* algorithm for path planning. The system, tested in simulations and closed-field experiments, achieved stable navigation to preset waypoints with sufficient accuracy. Additionally, the study designs a mechanical slide rail docking structure, validated through finite element analysis, showing minimal deformation under load and meeting stiffness and stability requirements. The work supports modular operations for combined air and ground transport modes.
This paper presents a mission-planning framework for autonomous marine vehicles conducting seabed surveys, focusing on safety, COLREG compliance, and operator transparency. It introduces a layered architecture for an onshore-supervised surface vessel that generates coverage plans from operator-defined polygons and produces structured reports for underwater vehicle inspections. The system includes a behavior-tree executive, a reinforcement-learning advisor for optimized survey edits, and a risk assessment layer for collision avoidance. The framework is validated through ongoing development of a surface vessel and remote operation center, using repeatable logs and status reporting.
This paper presents a 2-DOF tendon-driven manipulator for aquatic Autonomous Surface Vehicles (ASVs), such as the MallARD platform, to expand their operational range for both underwater and surface tasks. The design places actuators above the waterline, using wires to power the joints, allowing operation in high-radiation environments, handling heavy payloads, and resisting water exposure. A real-world prototype was built and tested, demonstrating feasibility with joint angle errors below 2.1° RMS. The inverse kinematics were solved analytically, and experimental results confirm its potential for underwater applications.
The paper introduces MAAT (Marine-Aware Adaptive Tracker), a multi-object tracking system designed for maritime environments to improve autonomous navigation of unmanned surface vehicles (USVs). It enhances ByteTrack by adding an adaptive Kalman filter and a density-aware association strategy to handle challenges like platform jitter and nonlinear object motion. Tested on the Singapore Maritime Dataset, MAAT achieves 44.37 MOTA and 43.857 IDF1 while running in real time at 41.4 FPS. The results show MAAT provides robust and accurate tracking for dynamic maritime conditions.
**Summary:** This paper introduces KDM-YOLO, a lightweight visual localization method for detecting steel surface defects, built on YOLOv10n. The approach improves feature extraction, context modeling, and multi-scale fusion using KWConv, C2f-DRB, and an MSAF module. Results show Precision, Recall, and mAP@50 of 91.0%, 93.9%, and 95.4%, with 3.29M parameters and 155.6 f/s inference speed.
128Ocean News & Technology2026-11-04
Ocius Technology signed a contract to verify its Bluebottle uncrewed surface vessels (USVs) for Japan’s Maritime Self-Defense Force (JMSDF). The solar, wind, and wave-powered USVs enhance maritime domain awareness with long-endurance, cost-effective surveillance above and below the ocean surface. The verification program strengthens collaboration between Australia and Japan on regional maritime security. The vessels will be built at Ocius Technology’s Sydney facility, following a $176 million Australian government investment in 2026 to expand Bluebottle production.
129Бабель2026-11-03
German company Stark is developing an unmanned ship designed to monitor Russian shadow fleet vessels. The project aims to enhance tracking capabilities for these often-sanctioned ships. No specific timeline or technical details about the vessel’s deployment were provided. The initiative reflects efforts to address maritime surveillance challenges.
This study developed an autonomous tracked robot for spraying in sheep houses, using a three-level navigation system with ROS to handle challenges like narrow aisles and animal blockages. Simulations showed high mapping accuracy (96.8% intersection-over-union) and precise localization (RMSE of 0.0299 m), while physical tests achieved near-complete coverage (99.8%) with minimal overlap. Spray deposition rates remained consistent across speeds (0.1985–0.2008 L/m²), though flow rate increased with speed. Field trials demonstrated reliable path tracking (RMSE 0.0376 m) and effective handling of animal blockages (55/56 events resolved automatically).
This paper presents a method using model predictive control to maintain stable hydroacoustic communication between an AUV and a support vessel or USV. The approach optimizes vessel positioning by predicting AUV trajectories, vessel dynamics, and hydroacoustic channel conditions to maximize signal-to-noise ratio (SNR). Simulations showed the method prevented communication loss and reduced unnecessary vessel movements compared to alternatives. The technique was validated as effective for reliable AUV operations in open sea conditions.
The SGR-YOLO model improves ship detection in complex maritime conditions by enhancing YOLOv8n with lightweight modifications. It integrates Sim-BiFPN for cross-scale feature fusion, GhostConv to reduce computational complexity, and RepViT to strengthen small-object detection. Tested on SeaShips and McShips datasets, it achieves 98.7% and 93.3% precision, respectively, while reducing parameters by up to 5.2M. The model maintains high performance in fog, occlusion, and long-distance scenarios, offering a real-time solution for resource-limited systems.
This paper presents a multi-agent GeoAI system for Arctic eco-navigation that balances vessel safety, efficiency, ecological protection, and community impacts. The system integrates operational, physical, ecological, and social criteria, using specialized agents to process geospatial data, generate multi-objective routes, and provide decision support. It explicitly accounts for sensitive habitats like Essential Fish Habitat and seal critical areas while keeping human oversight for value-based decisions. The framework aims to improve transparency and social responsibility in Arctic maritime routing.
134oceannews2026-11-02
OceanAlpha’s L42B uncrewed surface vessel (USV) received Approval in Principle (AiP) from Bureau Veritas on July 31, 2026, after its design and operations were reviewed against autonomous shipping guidelines. The 8.51-meter aluminum catamaran is built for offshore hydrographic and geophysical surveys, supporting missions up to eight days with a 1,500 km range. It features modular payloads for seabed mapping and combines diesel-electric propulsion with remote supervision capabilities. The AiP confirms compliance with defined operating conditions and includes recommendations for further development.
135ZAWYA2026-10-21Naval News
EDGE led a UAE Navy demonstration integrating sovereign autonomous systems with U.S. Navy participation, showcasing unmanned maritime capabilities. The event featured unmanned surface vessels (USVs) performing AI-driven reconnaissance and protection missions using advanced sensors and command-and-control systems. Technologies included TACTICA AI’s Naval Warfare C2 Centre, HADAF targeting system, and platforms like the 170 M-DETECTOR vessel and SKYSHIELD-N counter-drone system. The demonstration highlighted interoperability between allied forces and EDGE’s role in advancing autonomous maritime defense.
136Hydro International2026-10-20
Sentiq Defence, founded by retired Italian Navy commander Antonio J. Belfiore, has launched as an integrator of autonomous and remotely operated maritime systems for defense, security, and blue economy sectors. The company offers uncrewed surface vessels (USVs), underwater surveillance sonars, ROVs, and aerial drones, aiming to simplify integration and provide turnkey solutions with training and support. Based in Italy, Sentiq focuses on addressing fragmented markets by tailoring commercial-off-the-shelf (COTS) systems to customer needs under the principle "command your space." Its growth plan includes expanding partnerships, building a local integration hub, and enhancing training and support services.
137MarineLink2026-10-09
Robosys successfully validated its VOYAGER AI maritime autonomy software. The software is designed for autonomous vessel operations in the maritime industry. The validation confirms its readiness for real-world deployment. No specific details on testing methods or outcomes were provided.
138RICentral.com2026-09-04
A local company is advancing the development of autonomous vessels, positioning itself as a key player in the future of unmanned maritime technology. The initiative focuses on designing and implementing self-navigating ships for commercial and industrial use. The project aims to enhance efficiency and safety in marine operations. No specific technical details or partnerships were disclosed in the report.
This paper presents a transformer-based model for predicting future maritime radar frames using past X-band observations and vessel motion data from GNSS. The approach adapts the EchoPT framework, originally designed for simulated sonar, to real-world radar data from the MOANA dataset. The model's performance is tested against persistence and geometric warp benchmarks in both single-frame and autoregressive settings. The study also identifies failure modes tied to architectural and dataset limitations, suggesting areas for improvement.
140openalex2026-07-08Andrew Philip Single
This study advances a real-time acoustic system for autonomous underwater vehicles (AUVs) to detect and track surface vessels using a small hydrophone array. It improves prior work by handling AUV self-noise, multi-target tracking via a particle filter with DBSCAN clustering, and closed-loop motion control. Field tests in Norfolk and Claytor Lake validated the system, while simulations confirmed stable tracking of one or two acoustic sources under varying conditions. A rear suppression tool was found essential for reliable performance when AUV self-noise interfered with detection.
The ROBBE system is an autonomous robot designed for coating removal and decontamination of steel components in nuclear dismantling, using real-time 3D scanning, digital-twin trajectory planning, and ultra-high-pressure waterjet technology. It replaces manual labor, reducing radiation exposure and improving safety by operating in a closed cabin with optimized tool paths and process parameters. After 12 months of testing, ROBBE achieved over 90% decontamination on most parts in under 30 minutes, demonstrating reliability and efficiency. The system is now at technology readiness level 7–8 and is being expanded for broader use in nuclear decommissioning.
This paper introduces a unified vision-based control system for autonomous wheelchairs, merging corridor following and obstacle avoidance using a single monocular camera. A convolutional neural network (CNN) predicts steering commands directly from images for corridor navigation, while image-based visual servoing (IBVS) handles obstacle avoidance by regulating detected obstacles in the image space. A supervisory controller prioritizes safety by switching between the two methods as needed. Tested on a low-cost embedded platform, the system shows stable performance in varied conditions, combining learning-based and analytical approaches for assistive mobility.
This study explores how augmented reality (AR) can be integrated into maritime navigation to assist young navigators. Based on a review of existing research, AR improves situational awareness by displaying navigational data directly in the user’s field of view, reducing distractions and aiding performance in low-visibility or high-traffic scenarios. It also supports team decision-making through real-time, context-aware information sharing. However, its effectiveness depends on well-designed interfaces and proper training.
144openalex2026-04-21Alex Beaudin, Bjørn Andreas Kristiansen, Kristoffer Gryte, Corrado Chiatante +3 moreDOI ↗ (opens in a new tab) This paper introduces OLAhGP, a Gaussian process-based path planning method for robotic environmental monitoring that adapts to multi-modal sensing data while incorporating constraints. The approach optimizes future waypoints in a receding horizon framework, using the Gaussian process posterior to guide decisions. Tested on an autonomous surface vessel with oceanic algal bloom data, it outperforms existing methods by improving path informativeness and reducing misclassification errors in chlorophyll-rich waters. Results show better accuracy in bloom detection with the same number of samples.
VAK-Former is a semantic segmentation framework for autonomous Unmanned Surface Vessels (USVs) in maritime settings, using a Swin-L based Mask2Former architecture. It features a hierarchical vision transformer backbone, multi-scale pixel decoding, and query-based mask classification, optimized for the LaRS maritime dataset. The release includes reproducible training and evaluation scripts built on MMSegmentation, PyTorch 2.1.2, and Python 3.8. The code and associated manuscript are archived on Zenodo for long-term accessibility.
146gCaptain2026-11-03
German startup Stark Defence is developing a 24-meter uncrewed surface vessel, the Vanta 24, to monitor undersea cables and track Russia’s shadow fleet in European waters. The vessel, capable of carrying 50 metric tons and traveling at 40+ mph, is designed to be rapidly deployable and cost a fraction of manned frigates. Stark, backed by investors like Sequoia Capital and Founders Fund, aims to compete with established defense contractors by focusing on software-driven, uncrewed systems. The company has secured NATO contracts for drones and is expanding into maritime defense amid rising tensions in the Baltic and North Seas.
147Unmanned Systems Technology2026-11-02
Agilica advanced to the finals of EURONAVAL 2026 by demonstrating resilient maritime autonomy and ultra-wideband (UWB) positioning technology. The system showcased enhanced navigation and operational reliability for unmanned maritime vessels. This innovation highlights progress in autonomous maritime solutions. The finals event underscores its potential impact on naval and defense applications.
This paper introduces an Intelligent Runtime Safety Monitoring framework for autonomous multi-agent systems (MAS) used in applications like autonomous vehicles and robotics. The framework continuously monitors agent states, actions, and communications during operation, integrating safety verification, risk assessment, trust estimation, and adaptive corrections. It can respond to safety risks by allowing, warning, blocking, or escalating events for human intervention. The study uses the nuScenes and Waymo Open Motion datasets for evaluation, outlining data processing and analysis methods but does not yet present completed benchmark results.
This paper presents a low-cost autonomous vehicle navigation system using only a monocular camera. It integrates Canny edge detection, ArUco markers, and YOLOv4 for boundary detection, localization, and object detection, respectively. The combined approach was tested on a four-wheel electric vehicle, demonstrating effective navigation under varying visual conditions. Results show the system provides useful localization and obstacle detection without expensive sensors.
150Marine Insight2026-10-15
The US Navy successfully demonstrated a drone landing on a moving ship without GPS or radar, relying solely on onboard cameras and sensors. The system uses commercial-standard equipment and advanced software to recognize the ship’s visual cues and predict deck movement for smooth landings. This approach eliminates costly hardware modifications and can scale across various unmanned systems, from small drones to large aircraft. The technology is now ready for operational testing and may be integrated into Navy programs, including tactical resupply drones and combat aircraft.
This paper introduces a two-stage method for segmenting ship draft-mark regions and localizing waterlines in challenging maritime conditions like reflections, stains, and uneven lighting. The first stage uses Mean Shift filtering and contour statistics to isolate draft-mark regions, while the second stage applies adaptive color clustering and watershed segmentation to detect the waterline. Tested on 150 images, the approach reduced contour interference by over 50% in complex scenes and achieved 95.7% accuracy in region localization. The method outperformed alternatives like Canny and K-means in handling weak contrast and tilted viewpoints.
This paper introduces MetaVLM, a search-based metamorphic testing method to evaluate vision-language models (VLMs) for autonomous underwater robots (AURs). The approach uses NSGA-II to apply minimal image transformations that expose VLM prediction failures, testing models like BLIP and CLIP against a random search baseline. Results highlight the strengths and weaknesses of these VLMs in underwater robotic software. The findings provide insights for improving quality assurance in VLM-based systems.
The paper introduces Autonomous Science Nodes (ASNs), lightweight (~1 kg) microprobes deployed by a mothership to perform limited scientific measurements in extreme environments where conventional spacecraft struggle. These mechanically hardened nodes use spherical encapsulation, distributed loading, and high-energy deployment to survive harsh conditions. The study analyzes mass allocation, materials, power, communications, and redundancy, showing how ASNs enhance spatial coverage, resilience, and sampling capabilities. Applications demonstrate the flexibility of this architecture for extending mission reach beyond traditional spacecraft limitations.
This paper introduces a modular sensor platform for maritime autonomy, integrating RADAR, LiDAR, IMU, GNSS, AIS, cameras (RGB and LWIR), and weather sensors to improve environmental awareness and vessel tracking. The design addresses challenges like sensor installation constraints, unpredictable conditions (e.g., fog, sea clutter), and long-duration missions. A ROS2-based software framework supports data management, hardware-in-the-loop testing, and compatibility with existing systems. The platform aims to standardize data collection, enhancing reproducibility and comparability across research projects.
This paper introduces RMP-Nav, a reactive local planner for differential-drive robots that uses Riemannian Motion Policies and hierarchical wavemaps for autonomous exploration. The method reduces local-planning runtime to under 0.1 ms per call, significantly faster than the optimization-based FALCO planner, while maintaining comparable waypoint-completion performance in open environments like Campus and Forest. It also improves obstacle clearance and reduces jerk but shows moderate performance drops in constrained spaces like garages. The findings suggest reactive control is effective in geometry-heavy settings, though optimization-based approaches may still excel in topologically complex areas.
This paper introduces a safety assessment framework for autonomous vehicles by combining Fault Tree Analysis (FTA) and Bayesian Network (BN) to address both failures and functional insufficiencies, such as perception and sensor limitations. The approach enables qualitative and quantitative analysis, accounting for dynamic dependencies and uncertainties in real-world driving conditions. A case study on object detection failures identified weather and occlusion as major contributors, responsible for 45.76% and 58.72% of failures, respectively. The framework offers actionable insights for improving autonomous vehicle reliability under the Safety of the Intended Functionality (SOTIF) standard.
157Marine Insight2026-10-13
China Merchants Energy Shipping has partnered with RINA to investigate AI-driven technologies for ship operations. The collaboration aims to enhance efficiency and safety in maritime activities. The initiative will explore the use of artificial intelligence in optimizing vessel performance. The project was announced on October 13, 2026.
This study introduces a method to assess ship conflicts in confined inland waterways using AIS data by analyzing ship behavior and maneuvers. It filters geometry-induced turning behavior and links ship-domain models to observed avoidance actions, optimizing parameters for crossing and overtaking encounters. Analysis of 10,256 trajectories from the Yangtze River revealed distinct spatial patterns: crossing conflicts clustered in two regions, while overtaking conflicts spread along the main channel. The findings offer actionable insights for improving safety management in inland waterways.
This paper reviews technologies and challenges for autonomous Crew Transfer Vessels (CTVs) and Service Operation Vessels (SOVs) in offshore wind farms. It highlights gaps in current manual operations, limited validation of autonomous systems, and the potential for improved efficiency, safety, and cost reduction. Key advancements in sensors, automation, and risk assessment are discussed, along with operational hurdles. The authors outline research priorities to advance autonomous vessel development.
160crossref2026-09-01Irene Mejía Hurtado, Juan Camilo Angarita Noriega, Enric Cervera Mateu, Raúl Marín PradesDOI ↗ (opens in a new tab) This study compares Nav2 and EasyNav for LiDAR-inertial localization in land and maritime robots using a shared Livox MID-360 LiDAR and NVIDIA Jetson Orin Nano setup. Both frameworks process data with Fast-LIO2 for odometry and point cloud reconstruction, but Nav2 uses online 2D costmaps while EasyNav relies on a pre-built occupancy map. In port tests, EasyNav delivered more consistent positioning, whereas Nav2 showed higher sensitivity to local perception changes. The work highlights differences in how each framework handles perceptual data integration.
This paper reviews digital twin (DT) applications in marine robotics, covering AUVs, ROVs, USVs, underwater gliders, and emerging platforms. A structured search analyzed 107 studies, identifying varied architectures, technologies, and validation methods but no fully operational deployments. Key challenges include balancing model fidelity with computation, environmental uncertainties, and communication limitations. The authors suggest physics-informed learning and edge computing as potential solutions but highlight the need for sustained, interoperable physical-virtual integration.
This paper explores how generative and predictive AI models enhance autonomous robotic surgery. It outlines the evolution of AI-driven surgical systems, from perception to control, and examines key principles like real-time prediction, multimodal fusion, and simulation-to-real transfer. The study highlights applications in tasks like suturing, dissection, and navigation, demonstrating potential for semi- and fully autonomous operations. Future research directions include surgical foundation models and hybrid physics-AI systems.
163MSN2026-06-09New York Post
Autonomous hunter ships equipped with advanced technology were deployed off the coast of San Diego to monitor Chinese "ghost fleets." These unmanned vessels aim to track and gather intelligence on ships operating without proper identification or regulatory compliance. The initiative focuses on enhancing maritime surveillance and security. The effort targets concerns over illegal or covert maritime activities.
This paper presents a new method for visual geolocation in maritime environments using a 50° camera to match ground images with simulated horizons from a Digital Elevation Model. The approach employs a siamese network and compact visual descriptors to correlate images without relying on GNSS or radar. Tested over a 25 km² area, it achieves a mean localization accuracy of 38 meters and heading accuracy of up to 8°. The method enables GNSS-free navigation for autonomous maritime systems in unfamiliar or hostile environments.
165Naval Today2026-11-05
STARK unveiled the Vanta 24, a 24-meter uncrewed surface vessel (USV) designed for combat and maritime security in the Baltic Sea and Northern European waters. The USV can operate independently or alongside naval vessels, carrying up to 50 tonnes of payload, including anti-ship missiles, a 30 mm gun, and long-range radar, with speeds over 35 knots. It is built to endure harsh Baltic conditions and perform missions like escorting warships, monitoring suspicious shipping, and protecting underwater infrastructure. The vessel is part of STARK’s Vanta family, controllable via Minerva software, and aligns with NATO’s hybrid navy concept to enhance naval capabilities while reducing crew needs.
166Smart Maritime Network2026-11-04
Hyundai Heavy Industries has developed an AI system to inspect ship coatings. The technology aims to improve the accuracy and efficiency of coating quality assessments. This system is expected to enhance maintenance processes in the maritime industry. No specific performance metrics or implementation timelines were provided.
167해양레저신문2026-10-30
An AI-powered autonomous robot won an innovation award for its ability to detect and collect marine litter. The robot operates independently, identifying and retrieving ocean waste. The technology was recognized for its potential to address environmental pollution. The award highlights advancements in robotic solutions for marine conservation.
168아시아투데이2026-10-28
South Korea is introducing AI across marine and fisheries sectors to improve efficiency. The initiative includes deploying unmanned vessels for operations. AI will also be used to predict seafood supply and demand. The technology aims to modernize and optimize the entire industry.
This paper introduces a cross-modal semantic–geometric filtering method for LiDAR SLAM in dynamic water-surface environments, addressing issues like moving vessels and reflection-induced noise in USV mapping. The approach combines visual semantic data with LiDAR point confidence, using temporal association and adaptive attenuation to filter out interference while preserving static structures. Testing in a port-basin sequence showed an 87.18% reduction in vessel-related noise and a 77.98% decrease in reflection-affected points, alongside a 24.28% improvement in pose accuracy. The method balances noise suppression and structural preservation, enhancing trajectory consistency in dynamic conditions.
170데일리뉴스2026-10-17
The article highlights a 1.4 trillion won initiative in shipbuilding, emphasizing the critical role of "data connection" in integrating AI across the entire process. This includes linking design phases with robotic systems and autonomous navigation technologies. The focus is on creating seamless data flow to enhance efficiency and innovation in the industry. No specific methods or outcomes beyond the emphasis on data integration are detailed.
171데일리방산2026-10-16
A MARTAC unmanned surface vessel (USV) has been equipped with SEA.AI machine vision technology, enhancing its capabilities beyond standard camera systems. The upgrade aims to improve maritime situational awareness and autonomous navigation. The integration addresses limitations of cameras alone in detecting and identifying objects at sea. The system was highlighted in a report by 데일리방산.
172The Defence Blog2026-10-13
The U.S. Navy has equipped its drone boats with AI-powered vision systems. This upgrade enhances the boats' ability to detect and identify objects autonomously. The technology aims to improve navigation and operational efficiency. The initiative reflects ongoing advancements in naval automation.
173openalex2026-10-07Lojze Žust
This paper focuses on improving obstacle detection for unmanned surface vehicles (USVs) by addressing key challenges in visual perception. It introduces **SLR**, a training method reducing annotation effort by 20× while maintaining performance, and **WaSR-T**, a model using motion cues to filter water reflections. The study also presents **LaRS**, a large maritime dataset for panoptic segmentation, and **PanSR**, a new method enhancing small-object detection and crowded scene handling. These advancements aim to make USV navigation more robust and practical.
174The Australian2026-10-06
An ex-US Navy captain has developed the AI-powered ocean drone *Sea Archer*, designed to monitor and track adversaries in Australian waters. The autonomous vessel is prepared for deployment, leveraging artificial intelligence for surveillance operations. The project aims to enhance maritime security and intelligence-gathering capabilities. No specific adversaries or operational details were disclosed.
The paper introduces FuzzyCS-Proto, a federated learning framework for classifying underwater ship structural surface conditions, addressing challenges like non-IID data, image degradation, and class imbalance. It uses fuzzy client selection based on reliability metrics (validation performance, uncertainty, image quality, and class balance) and class-wise prototype learning to improve cross-client alignment. Tested on a dataset with five classes (corrosion, defect, marine growth, normal surface, and paint peel), FuzzyCS-Proto outperforms baselines like FedAvg and FedProx, achieving Macro-F1 scores of 0.6501, 0.6292, 0.6641, and 0.5074 across four federated settings. Ablation and sensitivity analyses confirm the method’s effectiveness and interpretability.
This paper introduces a cost-effective method for autonomous mobile robots to locate and dock with transport trolleys using a single LiDAR sensor. The system identifies trolley legs within predefined detection zones and enables docking without requiring precise robot positioning, utilizing an ESP32-based control system and ROS framework. Experiments confirmed successful trolley detection and docking when the robot was positioned in front of the target under controlled conditions. The approach aims to improve automated transport systems in warehouses and manufacturing facilities.
This paper introduces a quantitative method for analyzing gaze similarity in maritime operations to assess perceptual performance and situational awareness. The approach eliminates subjective judgment, improving accuracy and reproducibility in experimental settings. An experiment using simulated remote ship operations demonstrated the method’s feasibility by analyzing fixation data from navigational scenarios. The results suggest it can complement existing frameworks for evaluating maritime professionals' perceptual skills.
178Hellenic Shipping News2026-06-26
The Seawork 2026 conference highlighted advancements in autonomous maritime technology. Industry leaders discussed the latest developments and future trends in shipping innovation. Awards were presented to recognize outstanding contributions in the sector. The event emphasized the growing role of automation in maritime operations.
179openalex2026-06-04Grigorios Papanikolaou, Ioannis Kontopoulos, Giannis Spiliopoulos, Dimitris Zissis +1 moreDOI ↗ (opens in a new tab) This paper introduces a method to improve real-time object detection in autonomous maritime navigation by using GPS trajectory data. It modifies a YOLOv8-based detector with an early-exit mechanism that adjusts computation based on vessel motion, such as distance and speed. Frames with close or fast-converging vessels use the full model, while simpler scenes use a reduced network. Tests show the approach maintains accuracy while cutting inference time and computational costs.
This paper presents a framework for certifying the safety of autonomous inland vessels by integrating sensor availability, perception reliability, and adaptive control. It uses a pre-computed lookup table to manage degraded sensors, reducing computational load while ensuring predictable behavior. The approach combines redundancy, reliability validation, and risk-based control to support safe navigation under varying conditions. The method aims to meet certification requirements for autonomous systems.
The paper introduces an optimized REINFORCE algorithm using strategic gradients to improve visual navigation for autonomous driving robots. The enhanced model achieves 95.7% accuracy and 91.2% recall, outperforming comparative methods. In simulations, the algorithm reaches a 92.5% navigation success rate, exceeding traditional methods by 15.8%. Real-world tests show a 20.3% reduction in navigation decision time.
This study presents a field-tested system for floating-waste capture using an autonomous surface vessel (ASV) with camera-based perception and a deep reinforcement learning (DRL) controller. The system, trained entirely in simulation, was deployed on a real ASV, achieving centimeter-level accuracy in waste capture under various disturbance conditions. A two-stage simulation protocol and perception abstraction module helped evaluate the "sim-to-real" gap, revealing that actuation-model fidelity was the primary limitation. The approach was demonstrated in real-world search-and-capture tasks over areas up to 450 m², with lessons for improving reliability in sim-to-real transfer.
This study introduces an LSTM-MPC control framework to improve velocity stabilization in Autonomous Underwater Vehicles (AUVs) under unpredictable marine conditions. The approach uses an LSTM neural network to predict velocity and integrates it with Model Predictive Control (MPC) to optimize performance while accounting for actuator limits. An online learning mechanism allows real-time adaptation to disturbances and sensor noise without relying on analytical models. Experimental results show the method outperforms PID-based controls, reducing tracking errors in rotational axes while maintaining precision in translational movements.
The paper *BIND-USBL* introduces a cooperative localization system for Autonomous Underwater Vehicles (AUVs) in GPS-denied environments, where drift from inertial navigation is a major issue. The framework uses Autonomous Surface Vessels (ASVs) with Ultra-Short Baseline (USBL) acoustic positioning to provide intermittent fixes, reducing dead-reckoning errors. Key components include a multi-ASV formation model, a conflict-graph-based scheduler for efficient acoustic updates, and delayed fusion of USBL data with inertial measurements. Simulations demonstrate improved localization performance based on survey scale, acoustic coverage, and ASV-AUV team geometry.
This paper presents a 3D point cloud dataset collected using a Livox Mid-360 LiDAR sensor, targeting real-world vessel detection on inland lakes. It includes small fishing boats with detailed hull structures, water reflections, and varied environmental conditions, formatted for direct use with the OpenPCDet framework. The dataset contains raw point clouds, annotated 3D bounding boxes, and pre-processed metadata for training and validation. It supports research in 3D object detection, segmentation, and maritime perception for unmanned surface vehicles.
186Naval News2026-11-06
KONGSBERG has launched two new sonars, Aegir17 and Aegir57N, expanding its Aegir SSA sonar family for maritime security and navigation, with all three models now available for order. Aegir17 offers long-range, 360-degree surveillance, while Aegir57N combines navigation and security in a compact design for unmanned vessels and fixed installations. The company is also developing an AI-powered virtual sonar operator with Viam to automate underwater object detection. The Aegir SSA family leverages commercial sonar technology and AI to enhance subsea monitoring for diverse maritime applications.
187Marine Insight2026-10-16
The Taiwan Coast Guard Administration (CGA) received its first V-BAT drones from Shield AI, with twelve units to be delivered by late 2026. These vertical take-off and landing (VTOL) drones will enhance maritime surveillance, coastal patrols, and law enforcement by providing persistent Maritime Domain Awareness (MDA) without requiring runways. Shield AI and local partner AIDC will train CGA operators and establish a support system, while the drones’ 12+ hour endurance and electronic warfare resilience improve operational flexibility. The V-BATs were selected for their cost efficiency and ability to detect illegal activity in the Taiwan Strait.
188Unmanned Systems Technology2026-09-01
Tocaro Blue and Furuno have partnered to improve radar perception technology for autonomous marine systems. The collaboration aims to enhance navigation and safety in unmanned vessels. Their joint efforts focus on advancing radar capabilities for better environmental awareness. The initiative targets applications in maritime autonomy.
This paper addresses maritime object detection challenges for unmanned surface vessels (USVs), focusing on real-time processing of high-resolution images despite edge computing constraints. It introduces a method using Vision Mamba (ViM) backbones, which leverage State Space Models (SSMs) to efficiently handle long-range dependencies and high-resolution imagery through tokenization. The approach includes a custom Feature Pyramid Network with downsampling, SSM layers, and token pruning to reduce computational load. The solution outperforms existing methods like RT-DETR with ResNet50 in balancing performance and efficiency for maritime detection tasks.
This study introduces a method for detecting ship coating quality using portable Vis/NIR spectroscopy and machine learning. The approach combines spectral transformation (Nippy) with feature selection techniques (PSO and GA) and ensemble learning models. Results indicate that the proposed method outperforms single baseline models and traditional feature selection methods, achieving 99.33% accuracy with LDA when using Nippy and PSO. The ensemble models maintained stable performance across different preprocessing and feature selection stages.
191UK Defence Journal2026-10-29
Leonardo will integrate Shield AI’s Hivemind autonomy software into the Proteus uncrewed rotorcraft, developed with the Royal Navy and UK Defence Innovation, to enable mission-specific capabilities like anti-submarine warfare and logistics. The agreement allows Leonardo’s engineers to tailor and sustain the software independently, avoiding reliance on foreign suppliers. Proteus, a modular vertical take-off demonstrator, supports the Royal Navy’s 2040 aviation strategy by reducing crewed helicopter workloads. The collaboration aims to expand autonomous solutions across Leonardo’s rotary-wing portfolio.
The Mantis ROV Navigation 01 system was deployed on August 19, 2026, aboard the Mantis ROV, operated by CanPac Marine Services Inc. It records positional data such as latitude, longitude, speed, heading, pitch, roll, and depth for ships, buoys, and AUVs. The system was installed on a mobile platform, and its data were archived and made accessible via Ocean Networks Canada’s Oceans 3.0 infrastructure. Quality assurance and derived data products followed standard procedures.
193Marine News Magazine2026-10-13
MARTAC has integrated SEA.AI’s maritime machine vision technology into its unmanned surface vessels (USVs). The system uses AI and sensors to enhance real-time object detection and situational awareness at sea. This integration aims to improve navigation safety and operational efficiency for autonomous maritime operations. The technology was announced in a 2026 collaboration.
194PR Newswire2026-09-21
Simbe has deployed over 3,000 autonomous robots, making it the largest fleet of shelf-scanning robots in retail. These robots monitor inventory, track stock levels, and improve store efficiency. The expansion reflects growing adoption of automation in retail operations. No other company has reached this scale in autonomous shelf intelligence.
The Mantis ROV Navigation 01 system was deployed on July 24, 2026, aboard the Mantis ROV operated by CanPac Marine Services Inc. It records positional data such as latitude, longitude, speed, heading, pitch, roll, and depth for ships, buoys, and AUVs. The system was installed on a mobile platform, and its data is stored and accessible via Ocean Networks Canada’s Oceans 3.0 infrastructure. Quality assurance and derived data products follow standard procedures.
This study applies transfer learning to classify ship hull cleanliness using images, based on the ISO 8501-4 standard. Pre-trained convolutional neural networks (ResNet50, EfficientNetB0, MobileNetV2) and hybrid models with XGBoost were tested on binary (cleaned vs. uncleaned) and multi-class (WA1, WA2, WA2.5) tasks. Binary classification achieved high performance despite limited data, while multi-class classification proved difficult due to subtle class differences. The method supports automated inspection for underwater robotic cleaning systems.
This study introduces RCF-Free, a multi-modal 3D object detection framework for unmanned surface vehicles (USVs) that eliminates the need for precise sensor calibration. The method uses a Triple-Path Cross-View Fusion module, a Mobile Self-Attention Module, and a BEV-Point feature encoder to improve 3D perception in dynamic water environments. The authors also present two new water-related datasets, FloW-BEV and WaterScenes-BEV, achieving mAPBEV50 scores of 60.5% and 69.3%, respectively. The model demonstrates strong performance in both water and autonomous driving scenarios, with cross-domain generalization.
This study examines small-scale fishermen in Chellanam Harbour, Kerala, and their awareness of I-boat, an AI-driven autonomous vessel designed to improve navigation and ecosystem monitoring. Findings from interviews with 60 fishermen reveal limited and inconsistent knowledge of the technology. Key barriers include the lack of a Malayalam interface, inadequate infrastructure, and no formal training. The authors recommend localized, language-specific awareness programs led by the fisheries department to improve adoption.
The *ezTwin* framework enables rapid creation of hydrologic digital twins using minimal geospatial inputs like reservoir coordinates and river centerlines. It generates 3D geo-registered scenes in Unreal Engine with Cesium-based Google 3D Tiles, displaying live water levels and river bathymetry variations. The system integrates an unmanned surface vessel to collect real-time sonar depth data, refining the twin’s mesh and dynamically adjusting survey routes to reduce uncertainty. This approach simplifies deployment, supports flood scenario planning, and ensures real-time synchronization with physical conditions.
200台灣英文新聞2026-11-05
The article discusses how Ukraine’s unmanned boat warfare experience could inform Taiwan’s asymmetric defense strategies, though direct application is limited due to differing sea conditions, communication challenges, and defense environments in the Taiwan Strait. The DSET report suggests trilateral cooperation (Taiwan-U.S.-Ukraine or Taiwan-Japan-Ukraine) to combine Taiwan’s manufacturing strengths with Ukraine’s software and combat expertise, while addressing export controls and intellectual property concerns. While Ukraine’s low-cost, high-impact unmanned boats have proven effective in the Black Sea, Taiwan’s harsher maritime conditions and lack of combat experience may reduce their immediate effectiveness. The report emphasizes the need for localized testing, modular development, and cost-benefit analysis before scaling procurement.
201硬是要學 SOFT4FUN2026-11-03
Taiwan’s "Feng Jin Mo Gui Yu" unmanned surface vessel, developed by CSBC, debuted at the 2026 MATURE Taiwan expo. The 8.6-meter trimaran features stealth design, AI target recognition, and a top speed of 35+ knots, with capabilities for asymmetric warfare and port security. It has accompanied Taiwan’s Hai Kun submarine in sea trials and supports swarm operations of up to 50 vessels per control station. While still a prototype without government contracts, CSBC estimates potential annual production of 40–500 units.
202长沙晚报2026-11-02
A northern megacity is deploying unmanned boats and drones to create an integrated digital protection system covering air, land, sea, and space. The initiative aims to enhance surveillance and security across multiple environments. This system combines technologies for coordinated monitoring and response. The project was reported by 长沙晚报 in 2026.
203自由軍武頻道2026-11-02
The first "2026 Taiwan Marine Technology and Unmanned Vehicles Exhibition" showcased three unmanned boats at Tainan Port, resembling Taiwan-developed "Fenjin Modu Yu" models. The event previewed various unmanned systems, including high-speed surface vessels from MARTAC and STR8, as well as underwater vehicles like Awareocean’s *Taihe 1* and BeeX’s hovering AUV. Taiwan’s Vice President, Hsiao Bi-khim, attended, emphasizing the exhibition’s role in advancing marine tech, national security, and international collaboration. The displayed drones featured autonomous navigation, extended endurance, and stealth capabilities for military and research applications.
204Marine Insight2026-10-28
The U.S. Navy will equip five Arleigh Burke-class destroyers with Applied Intuition’s $25 million AI-powered Intelligent Perception System (IPS) to enhance vessel detection. The system integrates infrared sensors, cameras, and navigation feeds to automatically identify, track, and predict maritime threats in real time, reducing human error risks. IPS processes data onboard using edge computing and logs ship traffic digitally for faster analysis and fleet-wide intelligence sharing. It also supports machine learning model updates by transmitting sensor data to shore stations when bandwidth permits.
The paper introduces SURT-DETR, a Transformer-based framework designed to improve small object detection for unmanned surface vehicles (USVs) in maritime environments. It addresses challenges like scale variations and background interference using three innovations: a sea–sky line context-aware module (SLCAM), an uncertainty-guided small object decoder (UGSD), and an uncertainty-regularized NWD loss. Tested on the Pohang Canal dataset, SURT-DETR achieved a 79.5% mAP and 69.2% small-object AP, outperforming the RT-DETR baseline with minimal added parameters. Ablation studies confirmed the effectiveness of its components for practical USV perception.
206劳动观察2026-10-25
CCCC Shanghai Dredging and Dahua Technology demonstrated an unmanned boat collaborating with a dredger for offshore surveying at the Marine Expo. The innovation eliminates the need for separate traffic boats during mapping operations. Multiple technological advancements from the companies were showcased at the event. The development aims to improve efficiency in marine surveying tasks.
207国际船舶网2026-10-20
Jiangsu Ocean Shipping is researching and applying full-scenario adaptive embodied intelligent ships for inland rivers. The project focuses on developing ships capable of adapting to various operational conditions autonomously. This initiative aims to enhance efficiency and safety in inland waterway transportation. The research is highlighted as part of shipowner developments in the maritime sector.
208华人头条2026-10-17
A globally first embodied AI ship-painting robot was unveiled at the 2026 4th Fujian International Marine Equipment Expo. The robot, showcased as a breakthrough in maritime technology, is designed for automated ship coating applications. The event highlighted its potential to improve efficiency and safety in ship maintenance. No specific technical details or performance metrics were provided.
209华人头条2026-10-17
A globally first embodied AI ship-painting robot was unveiled at the 2026 4th Fujian International Marine Equipment Expo. The robot, showcased as a breakthrough in maritime technology, is designed for automated ship coating tasks. The event highlighted its potential to improve efficiency and safety in ship maintenance. No additional technical details or performance results were provided.
210QQ News2026-10-14
The US Navy has equipped unmanned boats with AI-powered systems to autonomously identify ships without human monitoring. These AI "eyes" enable the vessels to detect and recognize other ships independently. The technology aims to reduce reliance on manual surveillance. The initiative was reported by QQ News in October 2026.
211EDR Magazine - European Defence Review2026-10-09
Exail introduced Neptis, a maritime surveillance system based on distributed autonomy. The technology aims to enhance monitoring capabilities in naval operations. Neptis leverages autonomous systems for improved efficiency and coverage. The system was unveiled by the European Defence Review in October 2026.
212WorkBoat2026-10-06
Saab UK and Kraken have partnered to develop an uncrewed surface vessel (USV) designed for drone detection. The system aims to enhance maritime security by identifying and tracking aerial drones. The collaboration combines Saab’s defense expertise with Kraken’s advanced sensor technology. No specific deployment timeline or technical details were provided.
213手机新浪网2026-10-03
CSSC Exploration 01, a civilian marine floating test platform built by CSSC, is set to conduct sea trials. It will test shipborne unmanned helicopters, unmanned surface vehicles (USVs), and underwater vehicles. The platform will also perform engineering tests, including deck layout and electromagnetic compatibility for ships. The trials aim to validate various maritime unmanned systems and ship technologies.
This study examines Korea’s fragmented maritime history archives, which are scattered across institutions and difficult to access. It proposes an integrated digital platform featuring unified search, AI-enhanced document access, and sensor-based data (e.g., underwater sonar, 3D imaging) to preserve and analyze maritime cultural heritage. The platform aims to boost research, public engagement, and K-content development while addressing gaps in systematic archiving. Recent advances in marine technology and digital humanities highlight the need for a centralized system to improve accessibility and global dissemination.
This study presents a combined procedure using heavy-class ROVs and manned diving to enhance safety and efficiency in underwater search and rescue during large marine disasters. It analyzes tidal-current impacts and operational limits, finding ROV-SSDS integration allows longer diving windows than ROV-SCUBA due to SCUBA’s stricter current restrictions. The procedure incorporates ROV-mounted sensors for situational awareness, target localization, and safety monitoring. A classification framework was also developed to guide the selection of ROV-only, manned-diving-only, or integrated operations based on disaster severity, vessel needs, and depth.
Qi Li and Liming Tao (2026) developed V-MNet, a deep learning framework that simultaneously performs retinal vessel segmentation and multi-disease classification from fundus images. The model uses a vessel-guided mechanism to improve classification accuracy, achieving a Dice coefficient of 0.831 for segmentation and an average AUC of 0.978 for disease classification across four public datasets. It also incorporates Grad-CAM for visual interpretability, highlighting lesion regions to aid clinical decision-making. Ablation studies confirmed the effectiveness of its multi-task and vessel-guided design.
217Hellenic Shipping News2026-11-03
KONGSBERG expanded its Aegir SSA sonar family with two new models, Aegir17 and Aegir57N, now available for order alongside the existing Aegir57. Aegir17 offers long-range, 360-degree detection for larger platforms, while Aegir57N provides compact, high-definition navigation and security for unmanned vessels and fixed installations. The company also partnered with Viam to develop an AI-powered virtual sonar operator to automate underwater object detection. The Aegir SSA family integrates with broader maritime surveillance systems for enhanced subsea and surface monitoring.
This paper presents an AI-based framework for quickly evaluating the hydrodynamic performance of deformed marine structures like ships and offshore platforms. The method uses multi-view images to reconstruct a 3D model via the HunYuan3D model, followed by numerical simulations to assess stability and motion responses. Validation on a deformed ship showed geometric parameter errors under 3% and results consistent with known damage effects. The approach offers a low-cost, efficient, and reliable tool for emergency decision-making in maintenance and safety planning.
SLA-YOLO is a lightweight SAR ship detection model based on YOLO11n, designed to improve accuracy while reducing computational demands. It integrates three key components: LESCNet for enhanced convolution, C3k2-DAPE for dynamic feature weighting, and SA-LSCD for scale-specific predictions. The model achieves 97.3% and 90.8% mAP@0.5 on SSDD and HRSID datasets, respectively, with 34.1% fewer parameters and 42.9% fewer FLOPs than YOLO11n. It also shows strong transfer performance and small-ship detection improvements, reaching up to 66.8 FPS on an RTX 3090.
CEC-YOLO is a lightweight debris-detection algorithm based on YOLO11n, designed for complex water surfaces. It integrates existing feature-extraction and calibration modules in a task-specific structure, improving multi-scale debris representation while maintaining computational efficiency. The model achieved 88.9% mAP@0.50 at 175 frames per second on an RTX 3090, outperforming YOLO11n and other detectors in accuracy-speed balance. Validation on a public dataset confirmed its effectiveness beyond the self-built data, offering a deployable solution for water-cleaning vessels.
This study introduces an Attention U-Net model for segmenting marine biofouling in real-world conditions, enhancing traditional U-Net with spatial attention gates and squeeze-and-excitation modules. The model was trained and tested on 504 annotated images from ship hull surveys under varied environmental conditions. It achieved a validation Dice coefficient of 0.814 and a macro-accuracy of 0.689. The approach aims to improve automated inspection systems, reducing manual effort and increasing detection accuracy in maritime maintenance.
This paper introduces a control framework (DO-ANFSMC) for collaborative autonomous surface (ASV) and underwater (URV) vehicles in search-and-rescue missions, addressing environmental disturbances and obstacle avoidance. The method combines adaptive neuro-fuzzy systems, disturbance observers, and sliding-mode control, with parameters optimized via an improved grey wolf optimizer (IGWO-CT). Simulations with one ASV and two URVs demonstrated low yaw-rate tracking errors (0.00185–0.01915 rad/s) and reduced path lengths (68.171–79.399 m) while ensuring obstacle clearance. Results show improved trajectory tracking, disturbance rejection, and navigation efficiency compared to alternative approaches.
This paper examines the use of underwater robots to improve deep-sea marine cable inspection and maintenance. It evaluates ROVs, AUVs, and hybrid systems like UVMS, highlighting their roles in navigation, damage detection, and repairs. Advanced technologies such as vision-based systems, sonar, and multisensory fusion are discussed for enhancing accuracy in harsh conditions. Case studies demonstrate the effectiveness of these robotic platforms in real-world cable monitoring and repair tasks.
This study developed a low-cost, modular VOC sensor system for unmanned surface vehicles (USVs) to detect oil spills in marine environments. The system was tested in lab and outdoor settings, showing reliable detection of oil-contaminated conditions despite variability from airflow and ambient factors. Results confirmed the sensor’s ability to distinguish oil spills from background environments. The findings suggest potential for scalable, data-driven oil spill monitoring using USVs.
This paper introduces a "backseat driver" system for an autonomous underwater glider to passively track sperm whales using their echolocation clicks. The system, mounted on a modified SeaExplorer glider, uses a four-hydrophone array to detect, separate, and estimate the angle of whale sounds, adjusting the glider’s direction in real time. Field trials demonstrated the system’s ability to follow whales at distances over 100 meters with minimal noise impact. The design enables long-term passive acoustic monitoring, with open-source code and data available.
This paper presents a ship engine room monitoring system based on Field Control System (FCS) that is open, decentralized, and integrated using computer networks and fieldbuses. It introduces a network controller as a field data collector to standardize shared data and improve interaction interfaces for engine room monitoring. The system aims to enable comprehensive, intelligent management of ships while supporting seamless integration with shore-based information systems. The design addresses big data-era demands in shipping technology applications.
227Pulse 2.02026-11-06
Windward, a maritime AI company, acquired Azimut.ai to enhance its vessel monitoring capabilities using optical sensors. This expands Windward’s platform by adding ground- and ship-based visual intelligence, improving detection of vessels that evade traditional tracking signals like AIS. The technology creates digital vessel fingerprints without raw imagery, integrating with existing data to assess risks and support maritime security. The acquisition strengthens Windward’s proprietary data collection and aims to provide more accurate, independent verification of maritime activity.
228Breaking Defense2026-10-22
Uncrewed surface vessels (USVs) enhance maritime awareness by conducting intelligence, surveillance, and reconnaissance (ISR) missions continuously. These vessels provide real-time data to improve monitoring of maritime environments. Their persistent presence reduces gaps in surveillance coverage. The technology supports naval operations by increasing situational awareness without risking crewed assets.
229Hellenic Shipping News2026-10-18
The ABS North America Regional Committee Meeting highlighted AI, robotics, and nuclear energy as key technologies transforming the shipping industry into a data-driven sector. ABS Chairman John McDonald emphasized the need for industry support in adopting these innovations, including digital twins and autonomous systems, while ensuring safety and regulatory compliance. Discussions also covered U.S. maritime revitalization, AI-driven optimization, and market trends from experts like Microsoft’s Daryl Willis and Morgan Stanley’s Ellen Zentner. The meeting underscored collaboration among industry leaders to advance sustainable and commercially viable maritime solutions.
230TradingView2026-10-14
EXA reported a 21% organic growth in Q3 2026, driven primarily by strong performance in maritime robotics and navigation sectors. The company’s expansion in these areas contributed significantly to its overall quarterly results. No additional financial or operational details were provided.
SeaReTrack is a multi-object tracking framework designed for unmanned surface vehicles (USVs) to improve maritime perception. It addresses challenges like wave-induced camera motion, unstable detections, and target re-entry by combining motion-compensated association, HSV-based appearance verification, and boundary-aware identity recovery. Tested on the USVTrack benchmark, it achieved 64.35 HOTA and 80.09 IDF1 with 28 identity switches while maintaining real-time processing. The system also performed well in adverse conditions like rain, low light, and target reappearances.
232safety4sea2026-10-07
AI, automation, and connectivity may shape the future of the shipping industry. These technologies could enhance efficiency, safety, and operational capabilities. The article suggests they will play a key role in advancing maritime logistics. No specific methods or outcomes are detailed in the provided text.
The paper introduces Wiener-guided spectral line enhancement (WG-SLE), a trainable front end combining delayed-reference prediction with vessel recognition optimization. WG-SLE uses Wiener-initialized prototypes and stop-gradient blocks to enhance tonal cues while directing classification gradients. Tested on DeepShip and ShipsEar datasets, it improves accuracy by up to 6.77 percentage points over Audio Spectrogram Transformer baselines. Results demonstrate that delay-predictable spectral estimates provide complementary recognition information.
The paper introduces YOLO-MKV, an object detection algorithm for unmanned surface vehicles (USVs) designed to improve accuracy in complex marine environments. It combines a Multi-Scale Adaptive Kernel Convolution (MSAK) module with a Mixed Local-Channel Attention (MLCA) mechanism to enhance multi-scale feature extraction while maintaining low computational cost. The algorithm also integrates a high-resolution detection head and a Vision Transformer module, alongside a dynamic focusing mechanism using WIoU v3 loss to refine bounding box precision. Tests on the WSODD dataset show YOLO-MKV outperforms baseline models, achieving 82.0% mAP@0.5 and 47.3% mAP@0.5:0.95 with minimal parameters (2.67M) and computational demand (11.0 GFLOPs).
235Naval Today2026-10-27
The US Navy awarded Applied Intuition a contract to install an AI-powered vessel detection system, the Intelligent Perception System (IPS), on five Arleigh Burke-class destroyers. IPS uses shipboard cameras and infrared sensors to detect, track, and classify vessels in real time, enhancing situational awareness for bridge crews and combat centers. The system builds on earlier data collection efforts and supports over-the-air updates, operational data sharing, and future AI applications. It also generates machine-readable records to improve intelligence analysis and reduce reliance on manual logs.
This study introduces a two-stage neural network framework for controlling the swimming direction of an underactuated underwater walking robot with electromagnetic feet and propellers. The first network maps target directions to foot-tip positions using a mixture density model, while the second converts these positions into joint angles. Validation in simulation and in-air experiments showed mean direction tracking errors of 4.25° and 1.53°, respectively. The approach addresses challenges in non-unique inverse kinematics for the robot’s four-degree-of-freedom legs.
This paper introduces an **ℒ₁-model predictive control (MPC) framework** using a *hand position* reference point for underactuated surface vehicles, improving trajectory tracking under disturbances and model uncertainties. By redefining the control point ahead of the center of gravity (CoG), sway disturbances—unmatched in CoG-based dynamics—become matched uncertainties, allowing effective adaptive compensation. The approach combines MPC for constraint handling with **ℒ₁ adaptive control** for real-time disturbance rejection. Simulations and field tests show better performance compared to baseline methods.
This paper introduces an information-driven, multilevel path planning framework for unmanned surface vehicles (USVs) to improve environmental monitoring of uninhabited islands and reefs. The framework uses three algorithms: one for broad mapping of irregular areas, another for fine-tuned local mapping adjusted to tidal changes, and a third for energy-efficient ecological sampling despite ocean currents. Tests show the method outperforms existing approaches in path length, coverage accuracy, and energy use. The system enhances observation efficiency and reliability for long-term monitoring tasks.
239Marine Technology News2026-08-15
The article discusses advancements in radar processing for maritime autonomy, focusing on how raw radar data is transformed into actionable decisions for safe navigation. It highlights the role of radar systems in improving situational awareness and collision avoidance for autonomous ships. The piece emphasizes the importance of real-time data interpretation to enhance safety in maritime operations. No specific methods or results are detailed beyond the general focus on radar-based decision-making.
This paper examines modern approaches to training specialists in hydrometeorological and hydrographic support for navigation, emphasizing the need for updated education due to digitalization, stricter safety demands, and evolving maritime conditions. It highlights the shift toward an integrated, competence-based model combining hydrography, meteorology, oceanography, and geoinformation technologies, with a focus on digital tools, big data, and simulation training. The study compares Ukraine’s education system—strong in theory but needing more interdisciplinary and practical integration—with international standards. Key recommendations include adopting digital platforms, practice-oriented learning, and analytical skill development to improve decision-making in complex maritime environments.
241Marine Link2026-11-06
ShipIn Systems appointed Kunal Pathak as Vice President of Strategic Partnerships to strengthen collaborations with marine insurers, P&I clubs, and industry stakeholders. Pathak, a former Master Mariner with over 12 years at sea and experience in marine insurance and risk management, will focus on expanding the reach of ShipIn’s AI platform, FleetVision. His role aims to enhance loss prevention and risk assessment through real-time operational data, supporting closer ties between shipowners and insurers. FleetVision, deployed on over 1,300 vessels, provides continuous AI-driven insights to improve safety and operational performance.
This paper introduces a lightweight detector for unmanned surface vehicles to identify weak water-surface targets in real time. It uses frequency analysis to enhance features, focusing on mid-frequency energy unique to water-surface imagery, and employs adaptive localization supervision to improve detection accuracy. Tests on FloW and PoTATO datasets show improved performance in detecting small targets while maintaining real-time speed. Future work includes embedded deployment and more rigorous evaluation.
This study developed a custom dataset of life jacket images to improve maritime search and rescue (SAR) operations using computer vision. It tested YOLO models (versions 5–12) and found that advanced image preprocessing boosted detection accuracy in poor lighting. YOLOv10 performed best overall, balancing precision and speed (43.9 FPS on a Tesla T4 GPU). The approach aims to enhance real-time identification of individuals in distress at sea.
This paper presents a method for underwater scene reconstruction using a 3D rotating scanning sonar, RTK, and IMU systems on an unmanned surface vehicle (USV). The approach combines multi-sensor data fusion, image correlation calibration, and multi-view acoustic imaging to reconstruct water column and seabed scenes accurately. The system is cost-effective and does not require precise USV motion control. Lake bed imaging tests confirmed its high-precision seabed reconstruction capabilities.
245Naval Today2026-11-06
Naval Group and Abu Dhabi’s Technology Innovation Institute (TII) have partnered to develop and commercialize autonomous underwater vehicles (UUVs). The collaboration combines TII’s expertise in autonomous systems with Naval Group’s naval design and integration experience, focusing on surveillance, reconnaissance, and combat UUVs for both contested and non-contested waters. The agreement aims to advance the UAE’s maritime technology capabilities and strengthen industrial cooperation with France. Key efforts include integrating TII’s extra-large UUV with Naval Group’s unmanned combat underwater systems.
246Ocean News & Technology2026-11-05
Naval Group and Thales launched CortAIx France on November 5, 2026, to enhance artificial intelligence capabilities for naval applications. The initiative aims to develop advanced AI solutions tailored for defense and maritime security. CortAIx France will focus on improving decision-making, autonomy, and operational efficiency in naval systems. The collaboration combines expertise from both companies to strengthen France’s position in naval AI innovation.
247Naval Today2026-10-27
Exail launched the Marins M1, a compact inertial navigation system (INS) for small naval platforms like RHIBs, patrol boats, USVs, and UUVs. Designed for size, weight, power, and cost constraints, it uses fiber optic gyroscope (FOG) technology to maintain accuracy in GNSS-denied environments, with a position drift of 1 nautical mile after 3 hours. The system provides reliable position, heading, roll, and pitch data for smaller or uncrewed vessels. Exail aims to make high-performance navigation accessible to platforms with limited space and budget.
248헤럴드경제2026-10-21
The Korea Research Institute of Ships and Ocean Engineering (KRISO) showcased advanced ocean technologies, including AI-powered ships and underwater robots. These innovations aim to enhance maritime operations and exploration. The developments were presented as part of KRISO’s vision for future ocean technology. No specific technical details or results were provided in the announcement.
249Naval News2026-10-16
MARTAC unmanned surface vessels (USVs) now incorporate SEA.AI’s real-time machine vision technology. The system enhances situational awareness by detecting and classifying objects at sea. This integration aims to improve autonomous navigation and operational safety for maritime missions. The upgrade was announced in October 2026.
250Breakbulk.News2026-10-01
SEA.AI tested machine vision technology on uncrewed surface vessels (USVs) during the NATO REPMUS exercise. The deployment aimed to enhance maritime situational awareness and autonomous navigation capabilities. The exercise involved real-world operational scenarios to evaluate the system's performance. No specific outcomes or results were detailed in the report.
251Defence-point.gr2026-09-30
Shield AI and Kraken partnered to integrate Shield AI’s Hivemind software into Kraken’s USV K3 Scout unmanned surface vessel. The collaboration aims to enhance autonomous navigation and decision-making capabilities for maritime operations. The USV K3 Scout will leverage Hivemind’s AI-driven systems for improved mission performance. The partnership was announced on September 30, 2026.
This paper discusses the role of unmanned marine vehicles (UMVs), such as surface vehicles, underwater vehicles, and gliders, in ocean exploration, resource development, and environmental monitoring. It covers key technical aspects, including perception, planning, control, and swarm coordination. UMVs are highlighted as essential tools for maritime security and search-and-rescue operations. The work provides an overview of current advancements and applications in the field.
This paper introduces a framework to monitor maritime navigation integrity in fairways under GNSS jamming and spoofing threats. It combines AIS dead-reckoning with a GNSS RAIM baseline, using a two-state Hidden Markov Model (HMM) and a reinforcement-learning Deep Q-Network (DQN) to detect interference while maintaining low false alarms. The approach focuses on cross-track error (CTE) and protection bounds, improving detection rates—jamming (~93%) and spoofing (~99.5%)—compared to GNSS-only methods. Jammer/spoofer localization is also demonstrated, with accuracy ranging from meters to kilometers depending on geometry.
254openalex2026-07-21Yuan Gui, Hongchen Luo, Liqi Qu, Longyue Fu +1 more
The paper introduces **NaviAIS**, a standardized dataset for vessel trajectory prediction, addressing limitations in existing AIS datasets like inconsistent sampling, noise, and lack of structured navigational data. It provides unified temporal windows, local coordinate systems, vectorized lane priors, and rasterized navigable maps to improve reproducibility and environment-aware forecasting. The authors also propose **NaviLane**, a hierarchical framework that encodes trajectory-map interactions, generates multimodal predictions, and refines them using a consequence-aware evaluator. Experiments demonstrate NaviLane’s superior performance over baseline methods in both single-modal and multimodal prediction tasks.
The study compares Convolutional Neural Network (CNN) and Vision Transformer (ViT) architectures for detecting ships in maritime environments to improve security. It evaluates these AI-based systems for real-time object detection under varying conditions. The research focuses on designing and assessing intelligent detection models to enhance maritime surveillance. No specific performance results or methods beyond the comparison are detailed.
This paper explores adaptive navigation strategies for Autonomous Underwater Vehicles (AUVs) used in ocean exploration, monitoring, and military tasks. It highlights challenges like unpredictable ocean currents, moving obstacles, sensor limitations, and communication issues in dynamic marine settings. The study aims to improve AUV reliability in changing underwater conditions. No specific methods or results are detailed in the provided text.
257openalex2026-06-03Xueming Liu, Lin Li, Xiang Zhou, Tianjiang Hu +1 more
This paper presents a method for multiple unmanned surface vehicles (USVs) to circle a target without relying on external positioning systems. The USVs use onboard sensors and communication to measure distances and angles, employing two types of Kalman filters to estimate positions relative to the target and each other. A formation controller based on coupled oscillators ensures the system remains observable while maintaining a uniform circular formation. Simulations confirm the approach works effectively.
258openalex2026-06-02Stepan Konev
The paper introduces StandardE2E, a unified framework for end-to-end autonomous driving datasets. It standardizes preprocessing, enabling cross-dataset experimentation and combined training via a single PyTorch DataLoader. The framework simplifies adding new datasets by requiring only a mapping to a shared schema. It currently supports six datasets and is available as an open-source Python package.
This paper by A. Dhyani (2026) presents a framework for safe autonomous vessel navigation in inland waterways under uncertain and abnormal conditions. It introduces a robust system identification method for vessel maneuvering models, a Nonlinear Model Predictive Control (NMPC) system for constraint-compliant navigation, and a sensor Fault Detection and Isolation (FDI) scheme to monitor and diagnose sensor faults. The study also proposes a risk mitigation approach using Bayesian Belief Networks and Partially Observable Markov Decision Processes to handle failures like collisions or grounding. Validation includes full-scale trials, simulations, and case studies demonstrating improved prediction accuracy, control performance, and fault resilience.
The paper *SeaGuard-AI* introduces an adversarial-robust framework designed for reliable sea state estimation in autonomous marine vessels. Authored by Mrs. Kanaka Tulasi P. Reddi, it focuses on improving the accuracy and security of image-processing-based systems against adversarial attacks. The work is scheduled for publication on April 7, 2026. Additional unrelated topics, such as VLSI surveys and free paper downloads, are mentioned but not part of the core research.