Podcast

Sea Control 269 – USMC Commandant General David Berger On Force Design

2 Comments

By Jared Samuelson

While busy dramatically reshaping his service, Marine Corps Commandant General Berger managed to stop by the Sea Control virtual studio (the first podcast program to make it on his Commandant’s reading list.) In a wide-ranging interview, we discussed force design, expeditionary logistics, the Light Amphibious Warship, anti-submarine warfare, Navy-Marine Corps command and control arrangements and more.

Sea Control 269 – USMC Commandant General David Berger On Force Design

Links

4. Force Design 2030, by Gen. David H. Berger, March 26, 2020. 
5. “Bring Back the Seaplane,” by David Alman, War on the Rocks, July 1, 2020.
6. “Extend Air Wing Range with Seaplane Tankers,” by Second Lieutenant David Alman, U.S. Naval Institute Proceedings, May 2021.
7. “Marines Will Help Fight Submarines,” by Gen. David H. Berger, U.S. Naval Institute Proceedings, November 2020.
8. “EABO Needs a New Naval Command and Control Structure,” by Captain Nick Oltman, U.S. Naval Institute Proceedings, May 2019. 
9. “Marines Break Ground on New War Game Center,” by Todd South, Marine Corps Times, May 15, 2021.
10. “Redefine Readiness or Lose,” by Gen. Charles Q. Brown, Jr. and Gen. David H. Berger, War on the Rocks, March 15, 2021.

11. Commandant of the Marine Corps Professional Reading Program 2020.

Jared Samuelson is Executive Producer and Co-Host of the Sea Control podcast. Contact him at [email protected].

This episode was edited and produced by Keagan Ingersoll.
Emerging Tech Week

A Roadmap to Successful Sonar AI

2 Comments

Emerging Technologies Topic Week

By LT Andrew Pfau

Even as the private sector and academia have made rapid progress in the field of Artificial Intelligence (AI) and Machine Learning (ML), the Department of Defense (DoD) remains hamstrung by significant technical and policy challenges. Only a fraction of this civilian-driven progress can be applied to the AI and ML models and systems needed by the DoD; the uniquely military operational environments and modes of employment create unique development challenges for these potentially dominant systems. In order for ML systems to be successful once fielded, these issues must be considered now. The problems of dataset curation, data scarcity, updating models, and trust between humans and machines will challenge engineers in their efforts to create accurate, reliable, and relevant AI/ML systems.

Recent studies recognize these structural challenges. A GAO report found that only 38 percent of private sector research and development projects were aligned with DoD needs, while only 12 percent of projects could be categorized as AI or autonomy research.1 The National Security Commission on Artificial Intelligence’s Final Report also recognizes this gap, recommending more federal R&D funding for areas critical to advance technology, especially those that may not receive private sector investment.2 The sea services face particular challenges in adopting AI/ML technologies to their domains because private sector interest and investment in AI and autonomy at sea has been especially limited. One particular area that needs Navy-specific investment is that of ML systems for passive sonar systems, though the approach certainly has application to other ML systems.

Why Sonar is in Particular Need of Investment

Passive sonar systems are a critical component on many naval platforms today. Passive sonar listens for sounds emitted by ships or submarines and is the preferred tool of anti-submarine warfare, particularly for localizing and tracking targets. In contrast to active sonar, no signal is emitted, making it more covert and the method of choice for submarines to locate other vessels at sea. Passive sonar systems are used across the Navy in submarine, surface, and naval air assets, and in constant use during peace and war to locate and track adversary submarines. Because of this widespread use, any ML model for passive sonar systems would have a significant impact across the fleet and use on both manned and unmanned platforms. These models could easily integrate into traditional manned platforms to ease the cognitive load on human operators. They could also increase the autonomy of unmanned platforms, either surfaced or submerged, by giving these platforms the same abilities that manned platforms have to detect, track, and classify targets in passive sonar data.

Passive sonar, unlike technologies such as radar or LIDAR, lacks the dual use appeal that would spur high levels of private sector investment. While radar systems are used across the military and private sector for ground, naval, air, and space platforms, and active sonar has lucrative applications in the oil and gas industry, passive sonar is used almost exclusively by naval assets. This lack of incentive to invest in ML technologies related to sonar systems epitomizes the gap referred to by the NSC AI report. Recently, NORTHCOM has tested AI/ML systems to search through radar data for targets, a project that has received interest and participation from all 11 combatant commands and the DoD as a whole.3 Due to its niche uses, however, passive sonar ML systems cannot match this level of department wide investment and so demands strong advocacy within the Navy.

Dataset Curation

Artificial Intelligence and Machine Learning are often conflated and used interchangeably. Artificial Intelligence refers a field of computer science interested in creating machines that can behave with human-like abilities and can make decisions based on input data. In contrast, Machine Learning, a subset of the AI filed, refers to computer programs and algorithms that learn from repeated exposure to many examples, often millions, instead of operating based on explicit rules programmed by humans.4 The focus in this article is on topics specific to ML models and systems, which will be included as parts in a larger AI or autonomous system. For example, an ML model could classify ships from passive sonar data, this model would then feed information about those ships into an AI system that operates an Unmanned Underwater Vehicle (UUV). The AI would make decisions about how to steer the UUV based on data from the sonar ML model in addition to information about mission objectives, navigation, and other data.

Machine learning models must train on large volumes of data to produce accurate predictions. This data must be collected, labeled, and prepared for processing by the model. Data curation is a labor- and time-intensive task that is often viewed as an extra cost on ML projects since it must occur before any model can be trained, but this process should be seen as an integral part of ML model success. Researchers recently found that one of the most commonly used datasets in computer vision research, ImageNet, has approximately 6 percent of their images mislabeled 5. Another dataset, QuickDraw, had 10 percent of images mislabeled. Once the errors were corrected, model performance on the ImageNet dataset improved by 6 percent over a model trained on the original, uncorrected, dataset.5

For academic researchers, where the stakes of an error in a model are relatively low, this could be called a nuisance. However, ML models deployed on warships face greater consequences than those in research labs. A similar error, of 6 percent, in an ML model to classify warships would be far more consequential. The time and labor costs needed to correctly label data for use in ML model training needs to be factored into ML projects early. In order to make the creation of these datasets cost effective, automatic methods will be required to label data, and methods of expert human verification must ensure quality. Once a large enough dataset has been built up, costs will decrease. However, new data will still have to be continuously added to training datasets to ensure up to date examples are present in the training of models.

A passive acoustic dataset is much more than an audio recording: Where and when the data is collected, along with many other discrete factors, are also important and should be integrated into the dataset. Sonar data collected in one part of the ocean, or during a particular time of year, could be very different than other parts of the ocean or the same point in the ocean at a different time of year. Both the types of vessels encountered and the ocean environment will vary. Researchers at Brigham Young University demonstrated how variations in sound speed profiles can affect machine learning systems that operate on underwater acoustic data. They showed the effects of changing environmental conditions when attempting to classify seabed bottom type from a moving sound source, with variations in the ability of their ML model to provide correct classifications by up to 20 percent.6 Collecting data from all possible operating environments, at various times of the year, and labeling them appropriately will be critical to building robust datasets from which accurate ML models can be trained. Metadata, in the form of environmental conditions, sensor performance, sound propagation, and more must be incorporated during the data collection process. Engineers and researchers will be able to analyze metadata to understand where the data came from and what sensor or environmental conditions could be underrepresented or completely missing.

These challenges must be overcome in a cost-effective way to build datasets representative of real world operating environments and conditions.

Data Scarcity

Another challenge in the field of ML that has salience for sonar data are the challenges associated with very small, but important datasets. For an academic researcher, data scarcity may come about due to the prohibitive cost of experiments or rarity of events to collect data on, such as astronomical observations. For the DoN, these same challenges will occur in addition to DoN specific challenges. Unlike academia or the private sectors, stringent restrictions on classified data will limit who can use this data to train and develop models. How will an ML model be trained to recognize an adversary’s newest ship when there are only a few minutes of acoustic recording? Since machine learning models require large quantities of data, traditional training methods will not work or result in less effective models.

Data augmentation, replicating and modifying original data may be one answer to this problem. In computer vision research, data is augmented by rotating, flipping, or changing the color balance of an image. Since a car is still a car, even if the image of the car is rotated or inverted, a model will learn to recognize a car from many angles and in many environments. In acoustics research, data is augmented by adding in other sounds or changing the time scale or pitch of the original audio. From a few initial examples, a much larger dataset to train on can be created. However, these methods have not been extensively researched on passive sonar data. It is still unknown which methods of data augmentation will produce the best results for sonar models, and which could produce worse models. Further research into the best methods for data augmentation for underwater acoustics is required.

Another method used to generate training data is the use of models to create synthetic data. This method is used to create datasets to train voice recognition models. By using physical models, audio recordings can be simulated in rooms of various dimensions and materials, instead of trying to make recordings in every possible room configuration. Generating synthetic data for underwater audio is not as simple and will require more complex models and more compute power than models used for voice recognition. Researchers have experimented with generated synthetic underwater sounds using the ORCA sound propagation model.6 However, this research only simulated five discrete frequencies used in seabed classification work. A ship model for passive sonar data will require more frequencies, both discrete and broadband, to be simulated in order to produce synthetic acoustic data with enough fidelity to use in model training. The generation of realistic synthetic data will give system designers the ability to add targets with very few examples to a dataset.

The ability to augment existing data and create new data from synthetic models will create larger and more diverse datasets, leading to more robust and accurate ML models.

Building Trust between Humans and Machines

Machine learning models are good at telling a human what they know, which comes from the data they were trained on. They are not good at telling humans that they do not recognize an input or have never seen anything like it in training. This will be an issue if human operators are to develop trust in the ML models they will use. Telling an operator that it does not know, or the degree of confidence a model has in its answer, will be vital to building reliable human-machine teams. One method to building models with the ability to tell human operators that a sample is unknown is the use of Bayesian Neural Networks. Bayesian models can tell an operator how confident they are in a classification and even when the model does not know the answer. This falls under the field of explainable AI, AI systems that can tell a human how the system arrived at the classification or decision that is produced. In order to build trust between human operators and ML systems, a human will need some insight into how and why an ML system arrived at its output.

Ships at sea will encounter new ships, or ships that were not part of the model’s original training dataset. This will be a problem early in the use of these models, as datasets will initially be small and grow with the collection of more data. These models cannot fail easily and quickly, they must be able to distinguish between what is known and what is unknown. The DoN must consider how human operators will interact with these ML models at sea, not just model performance.

Model Updates

To build a great ML system, the models will have to be updated. New data will be collected and added to the training dataset to re-train the model so that it stays relevant. In these models, only certain model parameters are updated, not the design or structure of the model. These updates, like any other digital file can be measured in bytes. An important question for system designers to consider is how these updates will be distributed to fleet units and how often. One established model for this is the Acoustic- Rapid COTS Insertion (ARCI) program used in the US Navy’s Submarine Force. In the ARCI program, new hardware and software for sonar and fire control is built, tested, and deployed on a regular, two-year cycle.7 But two years may be too infrequent for ML systems that are capable of incorporating new data and models rapidly. The software industry employs a system of continuous deployment, in which engineers can push the latest model updates to their cloud-based systems instantly. This may work for some fleet units that have the network bandwidth to support over the air updates or that can return to base for physical transfer. Recognizing this gap, the Navy is currently seeking a system that can simultaneously refuel and transfer data, up to 2 terabytes, from a USV.8 This research proposal highlights the large volume of data will need to be moved, both on and off unmanned vessels. Other units, particularly submarines and UUVs, have far less communications bandwidth. If over-the-air updates to submarines or UUVs are desired, then more restrictions will be placed on model sizes to accommodate limited bandwidth. If models cannot be made small enough, updates will have to be brought to a unit in port and updated from a hard drive or other physical device.

Creating a good system for when and how to update these models will drive other system requirements. Engineers will need these requirements, such as size limitations on the model, ingestible data type, frequency of updates needed by the fleet, and how new data will be incorporated into model training before they start designing ML systems.

Conclusion

As recommended in the NSC AI report, the DoN must be ready to invest in technologies that are critical to future AI systems, but that are currently lacking in private sector interest. ML models for passive sonar, lacking both dual use appeal and broad uses across the DoD, clearly fits into this need. Specific investment is required to address several problems facing sonar ML systems, including dataset curation, data scarcity, model updates, and building trust between operators and systems. These challenges will require a combination of technical and policy solutions to solve them, and they must be solved in order to create successful ML systems. Addressing these challenges now, while projects are in a nascent stage, will lead to the development of more robust systems. These sonar ML systems will be a critical tool across a manned and unmanned fleet in anti-submarine warfare and the hunt for near-peer adversary submarines.

Lieutenant Andrew Pfau, USN, is a submariner serving as an instructor at the U.S. Naval Academy. He is a graduate of the Naval Postgraduate School and a recipient of the Rear Admiral Grace Murray Hopper Computer Science Award. The views and opinions expressed here are his own.

Endnotes

1. DiNapoli, T. J. (2020). Opportunities to Better Integrate Industry Independent Research and Development into DOD Planning. (GAO-20-578). Government Accountability Office.

2. National Security Commission on Artificial Intelligence (2021), Final Report.

3. Hitchens, T. (2021, July 15) NORTHCOM Head To Press DoD Leaders For AI Tools, Breaking Defense, https://breakingdefense.com/2021/07/exclusive-northcom-head-to-press-dod-leaders-for-ai-tools/

4. Denning, P., Lewis, T. Intelligence May Not be Computable. American Scientist. Nov-Dec 2019. http://denninginstitute.com/pjd/PUBS/amsci-2019-ai-hierachy.pdf

5. Hao, K. (2021, April 1) Error-riddled data sets are warping our sense of how good AI really is. MIT Technology Review. https://www.technologyreview.com/2021/04/01/1021619/ai-data-errors-warp-machine-learning-progress/

6. Neilsen et al (2021). Learning location and seabed type from a moving mid-frequency source. Journal of the Acoustical Society of America. (149). 692-705. https://doi.org/10.1121/10.0003361

7. DeLuca, P., Predd, J. B., Nixon, M., Blickstein, I., Button, R. W., Kallimani J. G., and Tierney, S. (2013) Lessons Learned from ARCI and SSDS in Assessing Aegis Program Transition to an Open-Architecture Model, (pp 79-84) RAND Corperation, https://www.jstor.org/stable/pdf/10.7249/j.ctt5hhsmj.15.pdf

8. Office of Naval Research, Automated Offboard Refueling and Data Transfer for Unmanned Surface Vehicles, BAA Announcement # N00014-16-S-BA09, https://www.globalsecurity.org/military/systems/ship/systems/oradts.htm

Featured Image: Sonar Technician (Surface) Seaman Daniel Kline performs passive acoustic analysis in the sonar control room aboard the guided-missile destroyer USS Ramage (DDG 61). (U.S. Navy photo by Mass Communication Specialist 2nd Class Jacob D. Moore/Released)

Emerging Tech Week

Red Cell Analysis of a Mobile Networked Control System Supporting a Ground Force

Leave a comment

Emerging Technologies Topic Week

By Larry Wigington, Ruriko Yoshida, and Doug Horner

The U.S. Department of Defense (DoD) will increasingly employ unmanned, autonomous systems to support ground force operations.1 However, if an opposing force can track these unmanned systems, or identify patters in the systems’ behavior, it may be able to determine the location and intended movements of the ground force. In this article, we present an adversarial, or “red cell,” analysis of prior Naval Postgraduate School (NPS) work with an Unmanned Vehicle (UxV) Networked Control System (NCS) to determine if multiple, distributed, heterogeneous unmanned systems behave in a manner that may compromise the operations security (OPSEC) of a ground force.

A UxV NCS consists of heterogeneous agents, which include unmanned and manned systems, where each node has communications, sensing, and mobility capabilities and constraints. Research conducted at NPS by Lowry2 and Wachlin3 focused on a centralized and distributed adaptive submodular optimization approach to find near-optimal, near-real-time, position solutions for agents in an NCS. Their work has produced methodologies that can find near-optimal position recommendations—which optimize over system parameters, such as sensing and communication—to increase mission effectiveness. However, over time these recommendations can create predictable paths that may reveal the location and intended movements of a ground force.

Determining if this is the case requires two assumptions; paramount is the ability for the adversary to observe and collect position data about the UxV NCS. Our red cell analysis is based on a “worst-case scenario,” using data from a real-world experiment in which a network of autonomous vehicles supported a ground force. In this step, we assume the adversary has perfect information about previous force disposition, system configurations, and access to detailed training data, including time-stamped geolocation information that links the ground force to a UxV NCS. Our second assumption is that opposing forces will use Artificial Intelligence and Machine Learning Techniques to identify patterns in the unmanned systems’ behavior.

The Scenario

In November of 2017, NPS designed and conducted a Multi-Thread Experiment (MTX) series on San Clemente Island, which brought together faculty and students from across NPS, including the Center for Autonomous Vehicle Research (CAVR), Center for Network Innovation and Experimentation (CENETIX), Consortium for Robotics and Unmanned Systems Education and Research (CRUSER), and Joint Interagency Field Experimentation Program (JIFX), as well as fleet sponsors such as Naval Special Warfare (NSW), Commander, U.S. Third Fleet (COMTHIRDFLT), and Naval Information Warfare Systems Command (NIWC).

The experiment provided a platform to test the collaboration of a UxV NCS with manned assets (Figure 1). The scenario was built around an NSW team tasked with landing on San Clemente Island to conduct a direct action mission on a known target. The NSW team was supported by the autonomous systems to provide Intelligence, Surveillance, andReconnaissance (ISR) support, transportation, and battle-space awareness. The MTX generated a reliable dataset to test the hypothesis that an NCS generates predictable paths that could compromise the OPSEC of a ground force. Additionally, the scenario provides the opportunity to conduct a more robust analysis, as the NCS dynamically positioned itself to support the ground force.

Figure 1: Two dimensional illustration of one scene from a Multi-Thread Experiment (MTX) series on San Clemente Island in November 2017, which was designed and conducted by NPS.

The Analysis

Our analysis of the NCS utilized machine learning techniques to predict each autonomous vehicle’s path and the ground force’s path and target during the MTX. We further developed and applied a novel framework to automate the analysis of the NCS-generated data from all aspects of data science, including supervised and unsupervised learning. This framework establishes a foundation to automate future red cell analyses of NCS data, which could be later incorporated into the NCS methodology to improve the OPSEC of supported ground forces. 

The framework consists of two independent models and a hierarchical model. The UxV Forecast Model conducts univariate time series analysis of the x- and y- coordinates, where the x and y values are generated for a localized grid coordinate system in a Northing Easting Down (NED) configuration and the point (0, 0) is based at lat-33.03191N, lon-118.60428W. This model builds twelve independent models and then forecasts each model a certain number of steps into the future. The output of the UxV Forecast Model creates the predicted locations of all NSC components.

The Ground Force Triangulation (GFT) model uses numeric regression techniques to calculate the ground force’s location as a function of UxV positions. The trained GFT model is then used in the time series regression model; the outputs of the GFT model are compared to the validation data to determine if the model is sufficient. The time series regression model combines the predictions from the UxV forecast model with the optimized model from the GFT model to generate the ground force’s predicted locations, which can then be compared against their actual positions.

The Results

The Unseen Data Set (UDS) was obtained from the research conducted by Lowry with a distributed submodularity. Lowry modified the data set, which we used for training our models to create a new ground force path, objective, and simulated UxV positions. We used the UDS to conduct our analysis from the perspective of an adversary that is observing autonomous vehicles and a U.S. Navy warship off its coast. The adversary previously obtained the MTX data sets and has two models which may be able to predict a ground force’s path, if one is present.

Figure 2: Time Series Neural Network Regression Model Results – Unseen Data Set. Left: UAV actual paths and predicted paths via time series models. Right: The NSW actual and predicted paths via the triangulation model. Click to expand.

The right-hand image of Figure 2 compares the NSW team’s predicted and actual paths. What is most evident in this image is that the predicted path and the actual path are nearly identical. Over the 24-minute prediction horizon, the mean path deviation is only 39 meters; what is more impressive is that the largest path deviation occurs in the middle of the prediction space and is only 75 meters. The predicted NSW locations that occur after this mid-point converge back towards the NSW team’s actual path.

Conclusion

Our analytical models correctly identified the ground force’s intended movements in both scenarios. The ground force’s predicted path deviated from the actual path by an average of only 39 meters. The implications of these results are far-reaching as DoD begins to focus on competing with near-peer adversaries in the Indo-Pacific Theater, and the Marine Corps identifies the need for reconnaissance and counter-reconnaissance capabilities when conducting operations within the “weapons engagement zone.” Future analyses of this kind will be a valuable component in planning and executing missions involving ground forces supported by mobile networked control systems.

Maj. Larry Wigington is an operations research analyst for U.S. Marine Corps’ Manpower and Reserve Affairs. He previously served as Adjutant for the 26th Marine Expeditionary Unit and G-1 Operations Officer for 2d Marine Division. He earned a master of systems analysis and master of science in operations analysis at the Naval Postgraduate School, and a bachelor of science in criminal justice at Troy University.

Dr. Ruriko Yoshida is an associate professor of operations research at the Naval Postgraduate School. She was previously an assistant and then associate professor at the University of Kentucky from 2006-2016, and a postdoctoral researcher at Duke University and the University of California, Berkeley, from 2004-2006. Dr. Yoshida received her PhD (2004) in mathematics from the University of California, Davis. https://nps.edu/faculty-profiles/-/cv/ryoshida 

Dr. Douglas Horner is a research assistant professor in the Department of Mechanical and Aerospace Engineering at the Naval Postgraduate School (NPS). He was a previously a naval special warfare officer (SEAL) from 1989-2001. A graduate of Boston University with a bachelor of arts in math and economics, Dr. Horner earned his PhD in computer science, master of science in applied mathematics, and master of arts in national security affairs at NPS. https://nps.edu/faculty-profiles/-/cv/dphorner 

References

[1] Defense Science Board. The role of autonomy in DoD systems. Technical report, Defense Science Board, Washington, DC, 2012. https://fas.org/irp/agency/dod/dsb/autonomy.pdf.

[2] Bryan Lowry. Distributed submodular optimization for a uxv networked control system. Master’s thesis, Department of Mechanical and Aerospace Engineering, Naval Postgraduate School, Monterey, CA, 2020.

[3] Noah Wachlin. Robust time-varying formation control with adaptive sub-modulairty. Master’s thesis, Department of Mechanical and Aerospace Engineering, Naval Postgraduate School, Monterey, CA, 2018.

Featured image: April 29, 2018 – U.S. Marine Corps 2nd Lt. Michael Francica, with Combat Logistics Battalion 8, Combat Logistics Regiment 2, 2nd Marine Logistics Group, pilots an InstantEye quadcopter during an operations check for Marine Corps Warfighting Laboratory as part of Integrated Training Exercise 3-18 on Marine Corps Air Ground Combat Center, Twentynine Palms, Calif. (Credit: U.S. Marine Corps)

Emerging Tech Week

Leviathan Wakes: China’s Growing Fleet of Autonomous Undersea Vehicles

Leave a comment

Emerging Technologies Topic Week

By Ryan Fedasiuk

Since 2018, Indonesian fishermen have regularly reeled in autonomous, glider-like vehicles operating as far south as the Java Sea—part of China’s longstanding undersea vehicle research program first declassified in 2021. Over the past decade, details have sporadically emerged about China’s unmanned (UUV) and autonomous undersea vehicle (AUV) projects, but questions linger about which kinds of vessels the Chinese defense industry may be developing, and how the People’s Liberation Army (PLA) might use them in a future conflict.

This article draws on a wide array of primary sources—including advertisements from defense companies, PLA writings and research papers, and information released by state-run research institutes—to illuminate China’s growing fleet of autonomous undersea vehicles. After profiling three major AUV research institutes, the article identifies potential applications of China’s growing fleet of AUVs and continued barriers to development.

China’s Big Three Undersea Vehicle Developers

As in many other Chinese technology industries, the state plays a leading role in undersea vehicle development. In 1986, Chinese Premier Zhao Ziyang initiated the State High-Tech Development Plan (863 Plan) to fund billions of dollars in applied technology development. In 1996, marine technologies were added to the program, adding further fuel to China’s emerging undersea vehicle industry. In particular, three government-sponsored research institutions form the backbone of AUV development in China. Each began undersea vehicle research in the 1980s, and has gone on to pioneer lines of AUVs still in use today:

Shenyang Institute of Automation. Part of the Chinese Academy of Sciences, SIA’s Computer Vision Group (机器人视觉组) is at the forefront of state-backed unmanned technology and autonomy research in China. In 1981, SIA developed the HR-01, China’s first remotely-piloted undersea vehicle. The institute went on to develop the “Explorer” (探索者) series of fully autonomous undersea vehicles in the 1990s and 2000s, with later variants capable of diving to 6,000 meters. Today, SIA specializes in developing prototypes of medium and large undersea vessels, including the Sea-Whale 2000 (海鲸2000) and Qianlong (潜龙; “Hidden Dragon”) series of AUVs.

China Shipbuilding Industry Corporation. As the world’s largest shipbuilding conglomerate, CSIC’s myriad research institutes have made significant contributions to China’s AUV research and development, particularly the 701, 702, 710, and 714 Research Institutes. Since 2000, CSIC has been responsible for developing the Haishen (海神; “Poseidon”) series of AUVs, in addition to a new line of autonomous undersea gliders, such as the Haiyi 1 (海翼一号; “Sea Wing”).

Harbin Engineering University. Originally called the Harbin Institute of Shipbuilding, HEU began developing a “Smart Water” (智水) series of AUVs in 1991, and has gradually expanded AUV testing in the South China Sea. The Smart Water series today comprises five variants of different sizes, but HEU has developed additional AUV models for undersea surveying and mapping, such as the Weilong (微龙; Microdragon) 1, 2, and 3. HEU is also home to China’s State Key Laboratory of Underwater Vehicle Technology (水下航行器信息技术重点实验室), which specializes in developing human-occupied, remotely-operated, and autonomous surface and deep-sea vehicles.

New Players in AUV Development

Beyond China’s big three AUV design centers, a growing number of research institutes and private enterprises are entering the Chinese AUV market. A document published in 2019 by the Chinese Society of Naval Architecture lists 159 undersea vehicle research projects under development at more than 40 Chinese universities—a significant increase over the 15 major universities that had constructed undersea vehicle research teams just four years prior. Another document prepared by Dr. Wu Jianguo, a professor at Hebei University of Science and Technology, shows that more than 48 universities and 45 enterprises in China host major UUV and AUV projects.

The CCP’s military-civil fusion (军民融合) development strategy is also facilitating a boom in China’s private-sector AUV industry. By promoting resource and information sharing between the military and private technology companies, China has seen some success in its efforts to accelerate AUV research, as several enterprises are developing new lines of AUVs independent of the big three research institutes. While it is too early to say whether they may become globally competitive with companies like the U.S.-based Bluefin Robotics or Norway’s Kongsberg, Chinese enterprises such as Xi’an Tianhe Haiphong Intelligent Technology (西安天和海防智能科技) and Startest Marine (星天海洋) seem to be emerging as China’s national champions of AUV systems and equipment. In 2017, China’s International Ocean Technology Exhibition in Qingdao attracted representatives from more than 500 enterprises working on AUV systems and components.

How the PLA Navy Might Use Autonomous Undersea Vehicles

While China’s AUV fleet still primarily consists of early-stage research experiments and prototypes, scientific research papers and theoretical writings by the PLA Navy (PLAN) indicate that it is primarily interested in using AUVs for marine surveying and reconnaissance, mine warfare and countermeasures, undersea cable inspection, and anti-submarine warfare. Each of these applications rely on different AUV models, and carry distinct implications and risks for the U.S. Navy and its partners in the Indo-Pacific.

Marine Surveying and Reconnaissance

The PLAN’s most mature application of AUVs is in marine surveying and reconnaissance. Chinese and American analysts have long assessed that the United States would retain a significant undersea advantage in a Taiwan Strait contingency, and the PLAN has rapidly expanded its manned diesel submarine force to compensate for this disadvantage. As early as 2013, the PLA commissioned construction of the Great Underwater Wall (水下长城), a network of hydrosonic sensors deployed at depths of 2,000 meters, which are designed to detect adversary undersea vehicles operating in the South China Sea. More recent research papers indicate that the PLAN may field groups of small and medium-sized AUVs for a similar purpose.

Portable and light models have so far formed the backbone of the PLAN’s emerging AUV fleet—in part due to battery constraints. Early AUV prototypes, such as CSIC’s Haishen 100 (海神100; “Poseidon 100”) and SIA’s Explorer (探索者), could dive to less than 1000 meters each; but modern variants have grown both in size and capability. Undersea gliders are a particularly promising innovation, and it is no surprise that so many have been found across the Indo-Pacific: Because of their improved range and endurance, gliders can more reliably detect undersea objects and periodically surface to transmit that information to ground stations and surface vessels. In the event of a crisis, analysts warn that autonomous gliders could be used to overcome China’s significant disadvantage in undersea warfare by detecting and tracking the locations of U.S. submarines even beyond the first island chain. Examples of such vehicles include SIA’s Haiyi (海翼; Sea Wing); Tianjin University’s Haiyan (海燕; Sea Swallow); and the Hai Xiang (海翔; Sea Flyer) developed by CSIC’s 702 Research Institute. Still, that gliders must surface in order to transmit information to one of three intelligence processing centers offers a point of vulnerability, and Chinese experts still believe the United States has an advantage in undersea surveying and mapping.

Figure 1. “Marine Environment Detection by Underwater Glider” Source: CSIC 714 Research Institute.

Mine Warfare and Countermeasures

Sea mines are a core tenet of Chinese naval doctrine and planning. As early as 2013, researchers at the China Engineering Science and Technology Forum publicly acknowledged the significance of UUVs in deploying mines and mine countermeasures (MCM); and military analysts have commented on the potential mine-laying capabilities of China’s HSU001 large-displacement UUV first unveiled in 2019. Today, the PLA may choose from more than 26 variants of floating and submerged mines designed to attack all manner of enemy ships and submarines.

Many of the PLA’s undersea mines are produced by CSIC’s 710 Research Institute—which also produces UUVs. Research published by the 710 Institute recommends developing AUVs and UUVs for minelaying and reconnaissance, but public information about the PLA’s current AUV models is scarce. One research paper published in August 2020 concludes that, “With the transformation of our navy’s strategy, the scope of anti-mine operations will extend to waters beyond the first island chain.” Some scholars believe that the PLAN could field more than 50 mines for each manned submersible—a feat that may eventually be possible for unmanned vehicles, too, as the Chinese AUV industry develops larger and larger vehicles. Examples of mid- and large-sized platforms include SIA’s Qianlong 1 and 2, and the Haishen series of UUVs developed by CSIC’s 701 Research Institute.

Figure 2. Haishen Series of AUVs. Source: Hebei University of Science and Technology.

Undersea Cable Inspection

Advances in unmanned vehicle research may also permit the PLAN to use AUVs to tap or sever undersea fiber-optic cables in a conflict, which concentrate near northern Taiwan. These cables are crucial not just for information dissemination in Taiwan, but also the trans-Pacific data exchanges that facilitate global internet access—including in some parts of the United States. HEU, for example, advertises AUVs for “underwater engineering investigation and maintenance,” including pipeline inspection and repair. Research papers and procurement records published by the PLA indicate that it may outfit undersea vehicles with robotic arms and sensors to interact with undersea cables. Some units have expressed an interest in procuring “SPICE,” a UUV produced by Kawasaki, which comes equipped with a robotic arm to repair undersea cables and pipelines.

Seabed Operations and Anti-Submarine Warfare

In the future, the Chinese military may use large AUV models for anti-submarine warfare (ASW) and operations near the seabed. Trends among PLAN research publications suggest that it is building signature libraries for undersea target detection and recognition, and several PLA units have awarded research contracts related to deep learning-based image recognition and target identification systems for undersea vehicles. In particular, students and researchers at HEU have pioneered an AI-based “seabed image mosaic system” for sonar image processing. Undersea target recognition systems like this could prove useful in autonomously identifying and interacting with sea floor infrastructure and other submarine vehicles.

Large AUVs may also be outfitted with weapons to engage adversary submarines. To date, the Chinese defense industry has produced few large AUV models suitable for ASW, although the number is growing. One notable example is the HSU-001, a large-displacement vehicle that sports side-scanning sonar arrays and a magnetic anomaly detector. Other medium and large AUV prototypes set records for depth and distance in 2020, including SIA’s Haidou 1 (海斗一号) and Sea-Whale 2000. Weighing in at more than 3.5 tons, China’s largest AUV so far is the Haishen 6000 (海神6000; “Poseidon 6000”), a 25 foot-long prototype developed by CSIC’s 701 Research Institute. The system is apparently capable of diving to depths of 6,000 meters, and comes equipped with “multiple detection devices such as an ultra-short baseline positioning system, an aircraft black box search sonar array, deep-sea side scanning sonar, an underwater camera, and forward-looking sonar.” Neither the HSU001 nor the Haishen 6000 appear to be outfitted with mine-laying rails or other weapon systems, but further advances in large AUV models and battery life could permit them to carry such payloads.

Figure 3. Haishen 6000, China’s Largest Known AUV. Source: CSIC 710 Research Institute.

Barriers to Chinese AUV Development

Despite its demonstrable progress in undersea detection and navigation, the Chinese AUV industry still faces significant technical and bureaucratic barriers to developing undersea platforms for military use.

The largest barriers are technical—particularly battery life. Among the small, medium, and large AUV models catalogued in this paper, most cannot operate for more than 24 hours before they require refueling or recharging, and PLA officers expect U.S. military AUVs to suffer from similar limitations. Moreover, artificial intelligence is still an emerging technology, and the target recognition systems used in modern autonomous vehicles are not robust enough to reliably detect undersea targets, let alone engage them in combat. Despite the PLA’s progress in testing AUVs, research from the Center for Security and Emerging Technology (CSET) concludes that “the state of the current technology, the complexity of antisubmarine warfare, and the sheer scale and physics-based challenges of undersea sensing and communications all suggest these systems have a long way to go.”

Bureaucratic inertia and a culture of central planning also constrain China’s progress in undersea vehicle development. On the one hand, the CCP’s continued focus on military-civil fusion does seem to be alleviating some barriers to AI development in the public sector. For example, a catalogue of ocean engineering projects published by the Tianjin Municipal Government advertises dozens of AUV and UUV models for both civilian and military use. But the fact remains that the big three state-backed research institutes—SIA, CSIC, and HEU—still dominate AUV development, and private-sector innovation in China has not yet reached its full potential.

Ryan Fedasiuk is a Research Analyst at Georgetown University’s Center for Security and Emerging Technology (CSET). His work focuses on military applications of emerging technologies, and on China’s efforts to acquire foreign technical information.

Featured Image: PLA HSU001 large-displacement UUV at Chinese Military Parade, October 2019. (Public domain)

Fostering the Discussion on Securing the Seas.