Underwater Robot Autonomous Manipulation 2026
Underwater Robot Autonomous Manipulation 2026
From teleoperated ROVs to AI-driven intervention AUVs, underwater autonomous manipulation is rapidly maturing. This dataset snapshot covers manipulation hardware, perception systems, control architectures, and emerging multi-robot approaches across 2014–2026.
Five Technical Domains Defining Underwater Autonomous Manipulation
Underwater robot autonomous manipulation spans five interrelated technical domains: manipulator hardware (rigid multi-link arms, soft robotic end-effectors, bioinspired tentacle mechanisms), perception for manipulation (3D laser scanning, stereo vision, point cloud processing, tactile sensing), vehicle-manipulator control (dynamic coupling compensation, model predictive control, visual servoing), simulation and validation infrastructure, and human-robot interfaces including VR-assisted teleoperation and shared autonomy.
The core challenge identified across retrieved records is that Intervention AUVs must execute manipulation under simultaneous perturbation from ocean currents, dynamic coupling between vehicle and arm, and degraded visual conditions. Publications in this dataset cluster heavily between 2018 and 2023, with patent filings extending to 2025–2026, indicating an actively maturing field transitioning from teleoperated systems toward fully autonomous intervention capabilities.
Chinese institutions account for the largest share of patents in this dataset, with key assignees including Harbin Engineering University, East China University of Technology, Shenyang Institute of Automation (Chinese Academy of Sciences), and Shanghai Maibu Technology. European institutions such as NTNU Norway, University of Florence, and the Spanish TWINBOT consortium are strongly represented in academic literature but less visible in patent filings within this dataset.
The innovation concentration pattern in this dataset is bimodal: Chinese institutions lead on patent filing volume and system-level engineering, while European and US academic groups lead on peer-reviewed algorithmic research. In retrieved records, only 2–3 patents directly cover intent-inferring, AI-mediated control arbitration for underwater manipulation, representing a significant IP white space in this dataset.
Patent and Literature Activity Across Technology Clusters
Retrieved records in this dataset span five technology clusters, with vision-based perception and soft robotics showing the densest literature activity and Chinese institutional assignees dominating system-level patent filings. The following charts summarize cluster distribution and temporal filing patterns.
Records per Technology Cluster — Retrieved Dataset
Vision-based perception and soft robotic manipulators each account for the largest share of records in this dataset, reflecting the centrality of perception and end-effector design challenges in retrieved literature.
↗ Click bars to explorePatent Filing Activity by Phase — Retrieved Records (2014–2026)
In this dataset, patent filings accelerated markedly in the 2022–2026 frontier phase, with Chinese institutional assignees contributing the majority of system-level patents in retrieved records during this period.
↗ Click bars to exploreKey Deployment Domains for Underwater Autonomous Manipulation
Retrieved records document underwater autonomous manipulation systems deployed or tested across deep-sea science, offshore infrastructure, aquaculture, and underwater archaeology. The following cards highlight named deployment contexts with specific data points from retrieved records.
Pacific & Santorini Hydrothermal Sites
A 2022 field study demonstrated autonomous ROV manipulation at Pacific and Santorini hydrothermal vent sites, using hardware-independent computer vision pipelines for biological sample collection. Operations were conducted with 7-DOF hydraulic manipulators aboard the SuBastian ROV and Nereid Under Ice hybrid ROV. The study established hardware-independent autonomous sample collection and return workflows in extreme underwater environments.
Deep-Sea ScienceDeep-Sea Biology, 2,224 m Depth
A 2018 study documented shipboard design and field deployment of custom 3D-printed soft robotic manipulators via ROV to a depth of 2,224 m for delicate biological specimen collection. The manipulators were fabricated aboard the research vessel and targeted fragile deep-sea organisms where rigid grippers would cause damage. This deployment established a benchmark for rapid-fabrication soft manipulator deployment at depth.
Deep-Sea ScienceShallow-Water Aquaculture Harvest, ~30 m
A 2019 study demonstrated an opposite-bending-and-extension soft robotic manipulator on a small AUV platform successfully harvesting sea cucumbers, sea urchins, and scallops at approximately 30 m depth. Inverse kinematics-controlled soft actuators enabled gentle grasping of irregularly shaped marine organisms. The 2026 US patent for an autonomous seafloor harvesting system with AI-coordinated drone swarms and underwater buffer stations extends this work toward commercial-scale deployment.
AquacultureARROWS Project Underwater Archaeology
The ARROWS project (2018) developed a heterogeneous AUV team for archaeological survey, combining autonomous underwater vehicles for survey, mapping, and manipulation tasks at historical underwater sites. A 2023 CN patent from Bewei Robot Technology (Shanghai) Co., Ltd. extended this to multi-robot coordinated detection, sampling, and excavation workflows on a collaborative underwater archaeology robot platform. These systems represent the application of autonomous manipulation to culturally sensitive fragile environments.
Underwater ArchaeologyLeading Patent Assignees in Underwater Robot Manipulation (Retrieved Records)
In this dataset, Chinese institutional assignees account for the largest share of system-level patents. Harbin Engineering University filed a compact UVMS architecture patent in 2018 (CN, active) and an autonomous docking control patent in 2025, while East China University of Technology (Nanchang Campus) filed path planning patents in both 2025 and 2026 in retrieved records.
Top Assignees by Patent Count — Underwater Robot Manipulation (Dataset Snapshot)
↗ Click bars to exploreHarbin Engineering University
Harbin Engineering University holds 2 patents in this dataset spanning 2018–2025. The 2018 CN active patent covers a compact underwater robot and robotic arm system with autonomous task execution and auto-return capability for diverse sea conditions. A 2025 filing extends their portfolio to autonomous docking control for persistent AUV operations.
China — CNMCS FREE ZONE
MCS FREE ZONE holds 2 patents in this dataset: a 2020 WO filing and a 2023 GB active patent, both covering an enhanced reality underwater maintenance system using a Virtual Reality Manipulator (VRM). The system enables an operator wearing a VR helmet and holding controllers to drive ROV manipulator arms, with an AI module translating controller movements to arm commands and integrating torque and force feedback for tasks such as bolt insertion.
United Kingdom — GBFrontier Signals in Underwater Autonomous Manipulation (2023–2026)
The most recent filings and publications in this dataset (2023–2026) reveal five emerging directions: digital twin-integrated manipulation planning, AI-augmented shared autonomy, autonomous docking as a mission enabler, multi-robot coordinated manipulation, and commercial-scale aquaculture and seafloor resource extraction.
Digital Twin-Integrated Manipulation Planning
A 2026 CN patent from Shanghai Maibu Technology builds a three-layer digital twin platform — incorporating prior information, dynamic environment, and real-time detection layers — to coordinate surface USVs and underwater ROVs for seafloor target search and path planning using synthetic aperture sonar. Digital twin methodology is explicitly enabling real-time adaptive manipulation planning beyond what static pre-programmed trajectories can achieve. This represents one of the most structurally advanced control architectures in retrieved records.
AI-Augmented Shared Autonomy and Intent Inference
The 2024 US pending patent from Toyota Technological Institute at Chicago covers shared autonomy for remote human-collaborative ROV manipulation, backed by NSF National Robotics Initiative and NASA funding. A 2023 paper on visual-aided shared control proposes an attraction-field-based arbitration mechanism that dynamically weights human command versus autonomous visual servo control based on inferred human intent. In this dataset, only 2–3 patents directly cover AI-mediated intent-inferring control arbitration for underwater manipulation, representing a significant IP white space.
Rigid Multi-Link Arms vs. Soft Robotic Manipulators for Underwater Autonomous Manipulation
Click any row to explore further.
| Dimension | Rigid Multi-Link Arm (UVMS) | Soft Robotic Manipulator |
|---|---|---|
| Primary Actuation | Hydraulic or electric motors; 7-DOF configurations documented (SuBastian ROV) | Soft bladder actuators, twisted and coiled polymer (TCP) artificial muscle, pneumatic pressure |
| Depth Independence | Performance varies with depth; hydraulic systems require pressure compensation | Force output explicitly decoupled from working depth (2020 depth-independent actuation study) |
| Max Demonstrated Depth | Deep-sea hydrothermal vent operations documented (Pacific, Santorini) | 2,224 m demonstrated (2018 shipboard 3D-printed soft gripper ROV deployment) |
| Grasping Suitability | Structured industrial objects, valves, cables, rigid samples | Delicate biological specimens, sea cucumbers, scallops, sea urchins (~30 m, 2019 study) |
| Control Complexity | Dynamic coupling compensation between vehicle and arm is the central challenge; requires UVMS control architectures | Inverse kinematics for soft arms; wireless control demonstrated (Kraken octopus robot, 2021) |
| Field Fabrication | Factory-built; not field-fabricable | Shipboard 3D printing demonstrated (2018); enables rapid field adaptation |
| IP Activity (Dataset) | Harbin Engineering University 2018 CN active patent; multiple Chinese institutional filings | Multiple literature records 2018–2022; fewer dedicated patents in this dataset |
| Primary Application Domain | Offshore energy infrastructure, cable maintenance, valve manipulation, deep-sea sampling | Marine biology collection, aquaculture harvesting, fragile specimen retrieval |
Frequently Asked Questions: Underwater Robot Autonomous Manipulation
Based on retrieved records, the field spans: (1) manipulator hardware including rigid multi-link arms, soft robotic end-effectors, and bioinspired tentacle mechanisms; (2) perception for manipulation including 3D laser scanning, stereo vision, point cloud processing, and tactile sensing; (3) vehicle-manipulator control including dynamic coupling compensation, model predictive control, and visual servoing; (4) simulation and validation infrastructure; and (5) human-robot interfaces including VR control, shared autonomy, and teleoperation assistance.
According to the 2022 survey in retrieved records, I-AUVs must execute manipulation under simultaneous perturbation from ocean currents, dynamic coupling between vehicle and arm, and degraded visual conditions including turbidity and low light. The gap between terrestrial and underwater manipulation autonomy is most acutely attributed to perception degradation — the absence of reliable depth sensors equivalent to terrestrial Kinect-style devices.
In this dataset, Chinese institutional assignees are the most active. Key assignees include Harbin Engineering University (UVMS system patent 2018, autonomous docking 2025), East China University of Technology Nanchang Campus (path planning patents 2025 and 2026), Shenyang Institute of Automation Chinese Academy of Sciences (docking device 2025), Shanghai Maibu Technology (USV-ROV cooperative path planning 2026), and MCS FREE ZONE (VR manipulator patents GB 2023, WO 2020). Toyota Technological Institute at Chicago holds a 2024 US pending patent for shared autonomy.
In retrieved records, soft robotic manipulators have been demonstrated at depths to 2,224 m in the 2018 shipboard 3D-printed soft gripper ROV deployment for delicate biological specimen collection. Separately, a 2019 study demonstrated an opposite-bending-and-extension soft robotic manipulator harvesting sea cucumbers, sea urchins, and scallops at approximately 30 m depth on a small AUV platform.
In this dataset, only 2–3 patents directly cover intent-inferring, AI-mediated control arbitration for underwater manipulation, despite this being identified as the most practical near-term control architecture. The 2024 US pending patent from Toyota Technological Institute at Chicago (backed by NSF and NASA funding) and a 2023 paper on attraction-field-based arbitration are the most notable records in this space, but the patent landscape remains sparse relative to the strategic value of the technology.
The most recent filings in this dataset include: a 2026 CN patent from East China University of Technology for marine environment underwater robot path planning; a 2026 CN pending patent from Shanghai Maibu Technology for USV and ROV cooperative seafloor target search using a three-layer digital twin platform; a 2026 US pending patent for a seafloor harvesting system with autonomous drone swarms including AI coordination and underwater buffer stations; and a 2025 CN active patent from Shenyang Institute of Automation for autonomous underwater robot docking device control.
Data and insights on this page are based on a limited patent and literature dataset and are for reference only. Figures may not represent the complete technology landscape.