Book a demo

Cut patent&paper research from weeks to hours with PatSnap Eureka AI!

Try now

AUV patent landscape 2026: AI and swarm tech trends

Autonomous Underwater Vehicle Technology Landscape 2026 — PatSnap Insights
Innovation Intelligence

Autonomous Underwater Vehicles have reached a strategic inflection point. Deep reinforcement learning is commoditising path planning, Chinese universities dominate the filing landscape, and cross-domain heterogeneous systems — pairing AUVs with surface vessels and drones — are emerging as the highest-value whitespace in the 2025–2026 patent record.

PatSnap Insights Team Innovation Intelligence Analysts 14 min read
Share
Reviewed by the PatSnap Insights editorial team ·

From Sonar to Swarms: The AUV Innovation Timeline

Autonomous Underwater Vehicle innovation has unfolded across three distinct phases between 2012 and 2026, each defined by a step-change in the underlying AI and systems architecture. The earliest filings established sensing foundations and basic autonomy; the most recent describe dense swarms coordinated by large language models and powered by hydrogen fuel cells.

~35
CN (China) AUV records in dataset
6+
Records from Harbin Engineering University alone
8+
Distinct RL-based AUV path planning patents
2012–2026
Publication date range in dataset

The Foundational Phase (2012–2017) established the core systems architecture. ENI S.P.A. filed one of the dataset’s earliest entries — an autonomous submarine system for 4D environmental monitoring — in 2012. Raytheon Company introduced seabed image correlation for navigation path replication in a 2017 PCT filing, while the Shenyang Institute of Automation addressed cost-efficient acoustic positioning through single-beacon combined navigation. These filings share a common characteristic: they treat the ocean environment as a known, if challenging, operational context.

The Development Phase (2018–2021) saw multiple institutions simultaneously pursue reinforcement learning-based path planning and multi-AUV coordination. Harbin Engineering University filed back-to-back reinforcement learning path planning patents in 2019 and extended the approach to a PCT/JP filing in 2021. Japan’s National Institute of Maritime, Port and Aviation Technology established simulation-based optimal control in an EP filing in 2018. Multi-AUV swarm platforms were prototyped by Zhejiang University and Harbin Engineering University in 2019–2020, signalling that single-vehicle autonomy was no longer the frontier.

The AUV patent dataset spans publications from 2012 to 2026, with a pronounced clustering of advanced AI-driven filings concentrated in the 2019–2026 window, reflecting the adoption of deep reinforcement learning for real-time trajectory optimisation.

The Acceleration Phase (2022–2026) shows convergence of large language model (LLM) interfaces, cross-domain heterogeneous teaming, energy-aware planning, and dense swarm motion choreography. Zhejiang University of Science and Technology filed two successive trajectory planning patents in 2025 leveraging PPO-IIFDS hybrid frameworks. Northwestern Polytechnical University filed a cross-domain underwater target search-and-track method in 2026 combining UAV, USV, and multi-AUV assets. ROVCO Limited published a 3D model-anchored subsea navigation system for commercial inspection work in 2025.

Figure 1 — AUV Patent Filing Volume by Phase, 2012–2026
AUV Patent Filing Volume by Innovation Phase 2012–2026 0 10 20 30 ~6 ~14 ~25 Foundational 2012–2017 Development 2018–2021 Acceleration 2022–2026 Foundational Development Acceleration
Approximate filing counts by phase, derived from the PatSnap dataset. The Acceleration Phase (2022–2026) accounts for the majority of advanced AI-driven AUV filings, reflecting the rapid adoption of deep reinforcement learning and multi-vehicle coordination architectures.

Four Technology Clusters Defining the AUV Patent Landscape

AUV patent innovation is organised around four interdependent capability pillars: autonomous path planning and motion control; integrated navigation through sensor fusion; multi-AUV swarm coordination; and docking, recovery, and endurance systems. Each cluster has a distinct maturity profile, competitive density, and IP differentiation opportunity.

Cluster 1: Reinforcement Learning and AI-Driven Path Planning

This is the most densely populated cluster in the dataset, with filings spanning 2019–2026. Methods range from Deep Q-Network (DQN) and Deep Deterministic Policy Gradient (DDPG) for single-vehicle mission control to Proximal Policy Optimization (PPO) combined with fluid-dynamic trajectory fields for 3D ocean environments. Harbin Engineering University’s 2021 JP filing defines a three-layer task-action-operation architecture, using DQN for online mission replanning and DDPG for low-level actuator control during tunnel inspection tasks. Zhejiang University of Science and Technology’s 2025 filing integrates real seabed digital elevation model (DEM) data and 3D ocean current vectors into an Improved Interfered Fluid Dynamic System (IIFDS), then optimises trajectory parameters via PPO with a 16-dimensional state space. Yanshan University’s 2024 filing constructs a lightweight neurodynamic network to exploit local real-time ocean current data without requiring global environmental maps — a significant practical advance for resource-constrained vehicles.

What is PPO-IIFDS in AUV path planning?

PPO-IIFDS is a hybrid method combining Proximal Policy Optimization (a reinforcement learning algorithm) with an Improved Interfered Fluid Dynamic System (a trajectory generation framework that models vehicle motion as fluid flow around obstacles). Applied to AUVs, it allows a 16-dimensional state space to be optimised in real ocean environments using actual seabed DEM and current vector data, rather than simplified simulation environments.

Cluster 2: Sensor Fusion and Acoustic Navigation

Navigation in GPS-denied underwater environments is addressed through combinations of Inertial Navigation System (INS), Doppler Velocity Log (DVL), Long Baseline (LBL) acoustic positioning, and image-based seabed correlation. Harbin Engineering University’s 2023 CN filing applies Kalman filtering to tightly fuse INS, DVL, LBL, and pressure sensor data, specifically addressing the failure mode where fewer than four transponder signals are available — a critical edge case in real deep-water deployments. Raytheon Company’s approach, filed in both WO (2017) and JP (2019) jurisdictions, generates complex seabed images during ingress and matches them against local images during egress using normalised cross-correlation coefficients (NCCCs) to enable path replication without pre-laid infrastructure. Standards bodies such as IEEE have published extensive guidance on underwater acoustic communication protocols that underpin these navigation architectures.

AUV navigation in GPS-denied underwater environments relies on combinations of Inertial Navigation System (INS), Doppler Velocity Log (DVL), Long Baseline (LBL) acoustic positioning, and image-based seabed correlation techniques, with Kalman filtering used to fuse these data streams and handle transponder signal failures.

Cluster 3: Multi-AUV Swarm Coordination and Cross-Domain Heterogeneous Systems

This cluster reflects the most significant growth area in recent filings. The Academy of Military Sciences, PLA filed a collaborative route planning method in 2020 that groups waypoints into neighbourhood clusters to reduce computational complexity of formation-level route search, with sub-variants optimising for shortest transit time and maximum residual energy. Northwestern Polytechnical University’s 2026 CN filing combines a single UAV for wide-area aerial search, a single USV acting as a communication relay, and multiple AUVs for underwater pursuit — a kinematically coupled heterogeneous system. Harbin Engineering University’s 2025 CN filing uses Dubins rolling-horizon homotopic path planning to jointly optimise navigation constraints, communication range, shared field-of-view, and individual maneuverability constraints for surface-underwater heterogeneous teams. French start-up ARKEOCEAN’s 2023 FR filing enables dynamic repositioning by synchronising AUV clocks with surface device clocks for time-of-flight acoustic positioning without pre-laid transponder arrays.

Explore the full AUV swarm coordination patent landscape with PatSnap Eureka’s AI-powered search.

Explore AUV Patents in PatSnap Eureka →

Cluster 4: Docking, Recovery, and Endurance Systems

A distinct set of filings addresses the practical deployment cycle — how AUVs are recovered, recharged, and re-deployed without returning to port. Hangzhou Dianzi University’s 2019 CN filing combines ultra-short baseline (USBL) acoustic positioning, machine vision, and a multi-degree-of-freedom robotic gripper on a USV to capture and inductively charge an AUV at sea. Zhejiang University City College Binjiang Innovation Center’s 2024 JP filing integrates a hydrogen fuel cell energy module, vector propulsion in the tail housing, and a hydraulic hovering mechanism to reduce ballast actuator dependency and extend operational endurance. The Shenyang Institute of Automation’s 2026 CN filing addresses energy-aware task segmentation for seabed-docked long-resident AUVs (LRAUVs), using CRC-verified task transmission and automatic residual-area identification to continue interrupted surveys without human intervention.

Figure 2 — AUV Technology Cluster Filing Density by Approach
AUV Technology Cluster Filing Density: Reinforcement Learning vs Sensor Fusion vs Swarm Coordination vs Endurance Systems 0 4 8 12 ~16 RL & AI Path Planning ~16 Sensor Fusion & Navigation ~8 Swarm & Cross-Domain ~10 Docking & Endurance ~5 Approximate number of patents in dataset
Reinforcement learning and AI-driven path planning is the most densely populated cluster, with approximately 16 distinct patents identified. Swarm and cross-domain coordination is the fastest-growing cluster in 2024–2026 filings.

Who Is Filing — and Where: The Geographic and Assignee Landscape

China dominates the AUV patent dataset with approximately 35 records, a filing volume that dwarfs every other jurisdiction. South Korea follows with approximately 20 records, though most are adjacent automotive or UAV technology rather than pure AUV. Japan contributes approximately 7 records including key academic and defense-linked filings, while Italy accounts for approximately 6 records dominated by ENI S.P.A., Scuola Superiore Sant’Anna, and Technology Innovation Institute.

In the PatSnap AUV patent dataset, China (CN) accounts for approximately 35 records — the largest share of any jurisdiction — followed by South Korea (KR) with approximately 20 records (predominantly adjacent UAV/automotive technology), Japan (JP) with approximately 7 records, and Italy (IT) with approximately 6 records.

At the assignee level, Harbin Engineering University is the single most active filer with 6+ records covering reinforcement learning path planning, INS/DVL/LBL navigation, swarm simulation, and USV-AUV cooperative detection. Northwestern Polytechnical University and its Shenzhen Research Institute hold 4 records covering swarm behaviour platforms, Petri net decision control, and cross-domain heterogeneous cooperative search. The Chinese Academy of Sciences’ Shenyang Institute of Automation holds 2 records covering single-beacon navigation and long-resident AUV path planning. Western commercial players — Raytheon, CGG Services SA, ENI S.P.A., Subsea 7, IBM, and ROVCO — appear with targeted filings in navigation, inspection, and resource extraction, reflecting commercially driven rather than foundational research intent. French start-up ARKEOCEAN and the University of the Aegean (Greece) represent niche innovation pockets in fleet coordination and aquaculture inspection respectively.

“Chinese universities and state-affiliated research institutes account for the large majority of AUV-specific records in this dataset. R&D strategists in Western markets should treat these institutions as the primary technical benchmark — and potential IP encumbrance risk — across path planning, swarm control, and sensor fusion sub-domains.”

The geographic concentration reflects broader patterns in maritime technology investment. According to WIPO‘s global innovation data, China has consistently ranked among the top patent-filing nations across robotics and autonomous systems sub-domains. The OECD‘s ocean economy reports similarly identify Asia-Pacific nations as leading investors in marine autonomous systems R&D, with significant government-directed funding channelled through university research institutes — a structural advantage reflected directly in the assignee concentration visible in this dataset.

Figure 3 — AUV Patent Records by Jurisdiction in Dataset
AUV Patent Records by Jurisdiction: China Dominates with ~35 Records in 2012–2026 Dataset 0 10 20 30 ~35 ~20 ~7 ~6 ~4 CN KR JP IT Other China S. Korea Japan Italy US/WO/FR/GB+
China accounts for approximately 35 AUV-specific records in the dataset — more than all other jurisdictions combined. Note: South Korea’s ~20 records include substantial adjacent automotive and UAV technology rather than pure AUV filings.

Five Emerging Directions in 2025–2026 AUV Patents

Five clear forward trajectories are identifiable from filings dated 2024–2026 in this dataset. Each represents a distinct technical bet on where AUV capability constraints will be resolved next — and where IP differentiation opportunity remains open.

1. LLM-Integrated AUV Control

Zhejiang University’s 2025 CN filing integrates a large language model with Nonlinear Model Predictive Control (NMPC) through ROS2, enabling natural-language mission specification and flexible control reconfiguration in complex experimental scenarios. This is the only LLM-AUV integration patent in the dataset and signals an emerging human-machine interaction paradigm. The practical implication is that mission operators without deep robotics expertise could specify complex multi-step underwater tasks in natural language — a significant accessibility shift for commercial AUV deployment.

2. Dense Swarm Motion Planning with Time-Space Coordination

Hunan University Wuxi Intelligent Control Research Institute’s 2026 CN filing addresses formation reconfiguration for large, tightly packed AUV clusters using time-space staggering rather than physical separation to prevent inter-vehicle collisions. This approach is architecturally distinct from prior formation control methods that rely on minimum separation distances — it enables higher vehicle density in constrained underwater spaces such as port approaches or pipeline corridors.

3. Energy-Aware Long-Resident Seabed Operations

The Shenyang Institute of Automation’s 2026 CN filing on long-resident AUV path planning and Harbin University of Science and Technology’s 2025 CN filing on SOC state-based AUV energy efficiency optimisation both treat battery state-of-charge as a first-class planning constraint. This represents a conceptual shift from purely geometric path optimisation to energy-lifecycle mission management — a necessary evolution as AUV missions extend from hours to days or weeks on seabed-docked platforms.

Key finding: Endurance is the binding constraint

The combination of hydrogen fuel cell patents (2024), SOC-aware energy optimisation (2025), and long-resident seabed docking systems (2026) indicates that operational endurance — not peak capability — is becoming the primary constraint limiting commercial AUV deployment. IP portfolios focused on energy management algorithms, wireless underwater charging, and alternative propulsion systems will carry increasing strategic value.

4. Cross-Domain Heterogeneous System Integration

Multiple 2025–2026 filings from Northwestern Polytechnical University, Harbin Engineering University, and Tsinghua University describe joint AUV-USV-UAV architectures. Tsinghua University’s 2025 CN filing maximises USV communication channel capacity to AUV clusters using multi-agent reinforcement learning (MAPPO/MADDPG), treating the surface layer as a dynamic relay mesh. The USV’s role as both acoustic communication bridge and potential power relay is a consistent architectural pattern across these filings, suggesting it will become a standard component of advanced AUV deployment systems. Research published by Nature on multi-agent robotic systems has highlighted cross-domain coordination as a key enabler for persistent ocean observation at scale.

5. Hydrogen Propulsion and Alternative Energy Sources

Zhejiang University City College Binjiang Innovation Center’s 2024 JP filing integrates a hydrogen fuel cell energy module with vector propulsion and a hydraulic hovering mechanism, signalling movement beyond lithium-ion batteries toward hydrogen fuel cells as a means of extending AUV endurance, particularly for deep-dive applications where frequent surface recovery is impractical. The IMO‘s ongoing work on alternative marine fuels provides a regulatory context that may accelerate commercial adoption of hydrogen propulsion systems across the broader maritime sector, including AUVs.

Track emerging AUV energy and propulsion patents in real time with PatSnap Eureka’s technology monitoring tools.

Monitor AUV Innovation in PatSnap Eureka →

Strategic Implications for R&D and IP Teams

The AUV patent landscape presents five concrete strategic signals for R&D planners, IP counsel, and technology investors. Each is derived directly from the filing patterns and assignee concentrations visible in the 2012–2026 dataset.

RL-Based Path Planning Is Approaching Prior Art Saturation

Reinforcement learning-based AUV path planning is represented by at least eight distinct patents from multiple Chinese universities alone, covering DQN, DDPG, PPO, neurodynamic networks, and homotopic planning. The core RL-AUV control layer is approaching crowded prior art density. Future IP differentiation will likely require specificity in reward function design, real ocean data integration, or hardware co-optimisation — not further variations on standard RL algorithms applied to simplified simulation environments. Patent applicants in this space should audit their claims against the existing prior art density before committing to broad filings.

Cross-Domain Heterogeneous Systems Are the Highest-Value Whitespace

The combination of AUV, USV, and UAV into tightly coupled heterogeneous teams — with USV acting as acoustic communication relay and power relay — is an emerging area with fewer filers and higher strategic leverage. ARKEOCEAN (FR), Kunming Ship Equipment Research Institute (CN), and Harbin Engineering University are early movers. This space warrants active monitoring and potential partnership or licensing strategy, particularly for Western commercial players seeking differentiated IP positions outside the densely filed RL path planning cluster.

Cross-domain heterogeneous AUV systems — combining AUVs with Unmanned Surface Vehicles (USVs) and Unmanned Aerial Vehicles (UAVs) — represent the highest-value IP whitespace in the 2025–2026 AUV patent dataset, with fewer filers and higher strategic leverage than the densely populated reinforcement learning path planning cluster.

Aquaculture and Ecological Inspection Are Under-Served

With only two filings explicitly targeting fish farm net inspection (University of the Aegean, GR, 2024) and marine transparency monitoring (Technology Innovation Institute, IT, 2025), the intersection of AUV robotics and blue economy aquaculture is an open field for application-specific product development and IP positioning. The University of the Aegean’s filing describes an AUV equipped with computer vision and multi-function tools — stapler, pliers, and grabber — to detect, document, and repair damage in fish farm net cages, suggesting that practical deployment requirements in this domain are well-defined but patent coverage remains thin.

Chinese Institutions Define the Technical Benchmark

Chinese universities and state-affiliated research institutes account for the large majority of advanced AUV filings in this dataset. Harbin Engineering University, Northwestern Polytechnical University, Zhejiang University, Tsinghua University, and the Shenyang Institute of Automation collectively span every major technical cluster. R&D strategists in Western markets should treat these institutions as the primary technical benchmark and potential IP encumbrance risk across path planning, swarm control, and sensor fusion sub-domains. Patent landscaping against this prior art base — using tools such as PatSnap’s patent analytics platform — is a necessary precursor to any significant AUV R&D investment decision.

Endurance Architecture Is the Emerging IP Battleground

The convergence of hydrogen fuel cell patents (2024), SOC-aware energy optimisation (2025), and long-resident seabed docking systems (2026) indicates that operational endurance is becoming the primary constraint limiting commercial AUV deployment. IP portfolios focused on energy management algorithms, wireless underwater charging, and alternative propulsion systems will carry increasing strategic value as the market shifts from demonstrating peak capability to sustaining persistent ocean presence. The PatSnap Insights blog covers related energy and propulsion technology landscapes in depth.

Frequently asked questions

Autonomous Underwater Vehicle patents — key questions answered

Still have questions? Let PatSnap Eureka answer them for you.

Ask PatSnap Eureka for a Deeper Answer →

References

  1. AUV Action Plan and Operation Control Method Based on Reinforcement Learning — Harbin Engineering University, 2021, JP
  2. Three-Dimensional Dynamic Trajectory Planning for Autonomous Underwater Robots for Real Ocean Environments — Zhejiang University of Science and Technology, 2025, CN
  3. PPO-IIFDS Based 3D Dynamic Trajectory Planning Method and System for Autonomous Underwater Robots — Zhejiang University of Science and Technology, 2025, CN
  4. Reinforcement Learning-Based AUV 3D Path Planning Method — Harbin Engineering University, 2019, CN
  5. Lightweight Online Path Planning Method for Autonomous Underwater Robots — Yanshan University, 2024, CN
  6. AUV Navigation Method and System Based on INS/DVL/LBL Tight Integration — Harbin Engineering University, 2023, CN
  7. Navigation System for Autonomous Underwater Vehicle Based on Coherence Map — Raytheon Company, 2017, WO
  8. AUV Single-Beacon Combined Navigation Method Based on TMA Technology — Shenyang Institute of Automation, Chinese Academy of Sciences, 2017, CN
  9. Route Setting Method for Underwater Vehicle — National Institute of Maritime, Port and Aviation Technology, 2018, EP
  10. Underwater Dynamic Single-Target Cross-Domain Cooperative Search and Tracking Method — Northwestern Polytechnical University, 2026, CN
  11. Robotic Vehicle Navigation System and Method — Rovco Limited, 2025, WO
  12. Assembly Comprising a Plurality of Autonomous Underwater Vehicles and Method of Repositioning — ARKEOCEAN, 2023, FR
  13. USV-AUV Formation 3D Cooperative Detection Path Planning — Harbin Engineering University, 2025, CN
  14. Hydrogen Energy Driven AUV Suitable for Submarine Diving — Zhejiang University City College Binjiang Innovation Center, 2024, JP
  15. Path Planning Method and System for Long-Resident Autonomous Underwater Robots — Shenyang Institute of Automation, Chinese Academy of Sciences, 2026, CN
  16. AUV Motion Control Method and System Based on LLM Human-Machine Interaction Framework — Zhejiang University, 2025, CN
  17. Autonomous Underwater Robot Dense Swarm Motion Planning Method, Device and System — Hunan University Wuxi Intelligent Control Research Institute, 2026, CN
  18. Underwater Fisheries Net Monitoring System — University of the Aegean, 2024, GR
  19. Underwater Robotic Fish School for Monitoring and Transparency — Technology Innovation Institute LLC, 2025, IT
  20. WIPO — World Intellectual Property Organization: Global Innovation Index and Patent Statistics
  21. OECD — Ocean Economy and Marine Autonomous Systems R&D Investment Reports
  22. IEEE — Standards and Publications on Underwater Acoustic Communication and Autonomous Marine Systems
  23. IMO — International Maritime Organization: Alternative Marine Fuels and Maritime Autonomous Surface Ships
  24. Nature — Multi-Agent Robotic Systems and Persistent Ocean Observation Research

All data and statistics in this article are sourced from the references above and from PatSnap‘s proprietary innovation intelligence platform. This landscape is derived from a targeted set of patent and literature records and represents a snapshot of innovation signals within this dataset only — it should not be interpreted as a comprehensive view of the full industry.

Your Agentic AI Partner
for Smarter Innovation

PatSnap fuses the world’s largest proprietary innovation dataset with cutting-edge AI to
supercharge R&D, IP strategy, materials science, and drug discovery.

Book a demo