The general definition of AI is the capability of a computer system to perform tasks that normally require human intelligence, such as visual perception, speech recognition and decision-making. Machine Learning (ML) is a subfield of Artificial Intelligence which attempts to endow computers with the capacity of learning from data, so that explicit programming is not necessary to perform a task. One of the most successful machine learning algorithms is deep learning (DL) that allow high-level abstraction from the data, and this is helpful for automatic features extraction and for pattern analysis/classification.
The transformative potential of AI on civilian and military systems and operations was realized early by many countries and led to their National plans for their development. AI has now heralded a new age of warfare which is having a multiplier effect in all domains of warfare including air, ground, sea, space and cyber and uses technology that is affordable and widely available.
The Pentagon outlined its first AI strategy in a report released in Feb 2019. The plan calls for accelerating the use of AI systems throughout the military, from intelligence-gathering operations to predicting maintenance problems in planes or ships. It urges the U.S. to advance such technology swiftly before other countries chip away at its technological advantage. “Other nations, particularly China and Russia, are making significant investments in AI for military purposes, including in applications that raise questions regarding international norms and human rights,” the report says.
AI is particularly useful for quickly and efficiently processing large volumes of data in order to obtain valuable information. AI can assist in culling and aggregating information from different datasets, as well as acquire and sum supersets of information from various sources. This advanced analysis enables military personnel to then recognize patterns and derive correlations.
AI techniques are being developed to enhance the accuracy of target recognition in complex combat environments. These techniques allow defense forces to gain an in-depth understanding of potential operation areas by analyzing reports, documents, news feeds, and other forms of unstructured information. Additionally, AI in target recognition systems improves the ability of these systems to identify the position of their targets.
The U.S. Navy, via its Office of Naval Research, has been pursuing AI applications, which could be used to help train and equip personnel, while also taking over some of the more mundane tasks on board warships. In this way AI could help reduce workload across the Navy, but also increase the speed and even quality of human decision-making. AI could speed up reaction time for defensive weapons, which could be increasingly critical as more advanced and potentially lethal weapons are being developed. Anti-aircraft and anti-missile systems could have reaction times that exceed the best human operator.
Likewise, AI and machine learning could help with naval operation decision making, which can be greatly impacted by foul weather. Currently, more than 75% of missions are severely disrupted by uncertain weather, and machine learning could reduce those numbers. Researchers at the University of Connecticut are now developing optimization and machine learning algorithms to help plan for such uncertainty. UConn’s work is being supported by a $1.5 million grant from the Office of Naval Research. The new algorithm incorporates multi-week meteorological and oceanographic forecasts using the Navy Earth System Prediction Capability, and from this data, AI could reduce weather-related uncertainty in mission planning.
AI on Warships
According a new study from research firm GlobalData, “AI in Aerospace and Defense,” the development of these technologies could present a long-term cost-cutting potential, as well as providing easy-to-understand analysis based on large datasets. The study noted that whilst fully autonomous ships powered by AI may not become the norm within the next 10 years, the technology will increasingly be used to aid decision-making in coming years.
“Advanced navies significantly invest in AI, computer and communication technologies in order to have larger and more capable autonomous vessels,” said William Davies, aerospace and defense associate analyst at GlobalData. “For instance, the US Navy converted two existing commercial fast supply vessels into unmanned surface vehicles (USVs) for its Ghost Fleet Overlord Surface program, which aims to inform and accelerate the Navy’s Large and Medium USV programs,” Davies told ClearanceJobs via an email. “Furthermore, on June 7, the US Department of Defense awarded a $44m contract to Austal USA to carry out the design, procurement, production implementation, and demonstration of autonomous capability in Expeditionary Fast Transport (EPF) vessel, USNS Apalachicola (T-EPF-13).” Further, the US DoD recently awarded a $44m contract to Austal USA to design and develop autonomous capability for the USNS Apalachicola expeditionary fast transport vessel.
The U.S. Navy isn’t alone in making a significant investment in AI. The UK Royal Navy is also exploring AI for naval ships under its Defence Science Technology Laboratory’s (Dstl) Intelligent Ship project. The navy tested the Startle and Sycoiea AI applications for the first time at sea onboard the HMS Dragon destroyer and HMS Lancaster Frigate against a supersonic missile threat. The applications were developed as part of the Above Water Systems programme led by researchers of Dstl.
“The UK invested £4m in 2020 for warship AI development projects, which will help warships to process data and provide crews with improved situational awareness,” added Davies. “Moreover, in 2017, China announced its Next Generation AI Development Plan, with a goal of becoming the world leader in the technology by 2030 – and in 2020 the country unveiled a ‘multi-purpose unmanned surface vessel,’ as well as reportedly developing AI-enabled submarines.”
“What we’re doing is using AI and ML to do that heavy lifting, get things down to actionable intelligence or real-world facts that we can put together in intelligent ways so that our warfighters can actually orient themselves to make their decisions sooner,” said Jim Tootell, PMAT’s chief software architect.
The same technology could then address issues of fleet maintenance. According to recent reports from the U.S. Government Accountability Office (GAO), the Navy’s four shipyards handled 75% of maintenance from 2015 to 2019, and that led to numerous delays. A number of factors including inadequate planning for resources, unplanned work during execution, direct yard costs and work stoppages, played a role – but the development of predictive analytics via machine learning and AI could help improve the efficiency at the facilities.
Additionally, the US Navy is looking at how it can automate system maintenance, Rear Adm. Lorin Selby, chief of naval research said. Artificial intelligence can be used to detect potential areas of failure in systems such as aircraft, surface vessels and submarines before they become problems. “It’s the oil-based, lube oil systems, the reduction gears, the propulsion systems, the turbines, the electrical plan — all of those things can be properly monitored” with AI, he said. Artificial intelligence will help the military prevent failures and predict when maintenance has to be done.
Simple Technologies Solutions (STS), Google Cloud, and the Navy are working on an initiative to examine how artificial intelligence and machine learning can be used to identify ship corrosion. STS will deploy drones that will gather images in the first phase of the effort and will train Google Cloud AutoML to recognize signs of corrosion. PMAT is developing an AI software platform – X-CAP – for both the Navy and Coast Guard that is intended to help analyze data to enable warfighters to make decisions faster.
Verdict has conducted a poll to assess the time it would take for AI to be integrated significantly into naval vessels. Analysis of the poll results shows that a majority 75% of the respondents expect the significant integration of AI to take not more than ten years. While 43% expect it to take less than five years, 32% anticipate it to happen over ten years or lesser.
The world’s leading navies are increasingly deploying AI to enable warships to process data and provide enhanced situational awareness to the crew. Although the application of AI in the navy is still in the nascent stage, studies and programmes are underway to test the capabilities of the technology. The US Naval Surface Warfare Center, for example, is developing SWARM-Tac, which provides naval ships with situation awareness including the number of weapons onboard and the number of attackers. It provides data on the probability of success of a chosen solution on how to evade an enemy or destroy swarm of enemy boats. An at-sea test onboard a naval ship delivered significant results.
A warship at sea requires awareness of the ships in its vicinity in order to operate safely. At times, this can be a daunting task, even when equipped with multiple shipboard systems. Current warships with helicopter capabilities have shipboard systems that provide a partial surface tactical picture. Surface tactical picture comprises of information about ships around a warship displayed in a combat system to help it make tactical decisions. Forward-Looking Infrared (FLIR) imaging systems aboard most warships permit visual identification of all contacts within 12 nautical miles (nm). Some other onboard systems, such as radar systems, enable crews to make relatively accurate deductions within 30 nm.
Since 2004, the implementation of the Automatic Information System (AIS) by the International Maritime Organization (2002) allows crews to recognize many contacts within 50 nm by direct signal reception, simplifying and greatly enriching the surface tactical picture, although its range continuously varies with the propagation conditions of the atmosphere, and deception is possible without additional confirmation gained, for example, by an air asset.
If naval air assets are available, a Tactical Action Officer (TAO) directs them to gain additional information about as many surface Contacts of Interest (COIs) as possible. Warships use air assets—typically a naval helicopter or an unmanned aerial vehicle (UAV)—to complete the surface tactical picture. A surface search flight of any organic air asset—belonging to a destroyer or frigate—is typically conducted twice a day during peacetime operations. In a pre-flight briefing, the TAO, helped by the Air Asset or Helicopter Controller (HCO), establishes the intended flight pattern to be carried out by the pilots, as well as search priorities. Because a surface picture is dynamic, priorities sometimes change after the air asset is underway.
These air asset routes can be inefficient because there are no shipboard systems to aid with route planning. Additional complications include COIs moving during a route, and some COIs being more important to visit than others.
Alvaro Herraiz Solla of of Naval Postgraduate School has formulated and implemented two Optimal Routing of Coordinated Aircraft (ORCA) Integer Linear Programs (ILP) to plan air asset routes that visit as many prioritized COIs as possible in a fixed time horizon.
Dynamic frequency allocation in communications and electronic warfare plans
Development and acquisition of naval communication, data, and radar systems for ships is an almost entirely modular process. For this reason, virtually all existing systems have separate controllers, antennas, and transmitters.
Antennas for radio frequency (RF) systems on warships are typically designed for a single purpose. Individual antennas are normally optimized for frequency, radiation pattern, polarity, and power requirements. Additionally, warships’ superstructures are cluttered with single-purpose antennas. The superstructures of ships are metal and antenna transmission patterns change because of the presence of metal and other antennas. This typically results in three undesirable conditions. First, antenna patterns are unpredictable for both transmission and reception. Antennas mounted on metal structures are subject to interference from that same structure. In some cases, null zones are created where transmission or reception is not possible or severely degraded. Unobscured space on a warship’s superstructure is limited.
However, future systems could use existing planar antennas that operate across a range of frequencies and create a variety of complex waveforms, eliminating the need to develop separate antennas and transmitters. The Integrated Topside (InTop) joint Navy industry open architecture study published in 2010 described the need for an integrated sensor and communication system that is modular, scalable, and capable of performing multiple functions. Such a system requires a scheduling and frequency deconfliction tool that is capable of representing the current antenna configuration and matches those capabilities with requests for frequency space and time.
Steven J. Fischbach of Naval Postgraduate School has developed SPECTRA, an integer linear program that can prioritize and optimize the scheduling of available antennas to deconflict time, frequencies, systems and capabilities. It can be uniquely tailored to any platform including naval warships, aircraft, and ground sites.
Automated decision aid
Emerging threats to naval ships include lower-cost, unsophisticated systems that can overwhelm ship defensive capabilities when used in large numbers across multiple domains (air, surface and subsurface). A defensive system capable of countering swarming threats across multiple domains will be essential for naval forces that anticipate fighting in contested, high-threat environments in the future.
Pairing naval weapons to swarming threats in multiple domains with multiple defensive systems is an extremely difficult task that is currently performed manually. For the scenarios envisioned, the expected number of possible response options will involve numerous possible pairings and millions of unique order-of-engagements.
Laird, Christopher L. Monterey of Naval Postgraduate School has developed an automated decision aid capable of generating defensive engagement profiles for use in naval shipboard defense. It allows the efficient pairing of multiple defensive weapon systems to several incoming threats operating in multiple domains by providing the operator with recommended weapon-target pairings based on current defensive capabilities and threat profiles.
The model consists of a pre-processing algorithm and a reward-based mixed-integer programming model that takes as inputs the available defensive weapon system capabilities and incoming target information and outputs a recommended engagement profile. Recommended weapon-target pairings are based on the priority of the threat, the time available to engage it, and the probability of successfully countering it.
Artificial intelligence system helps Navy select the best tactics for ship defense
At present human operators would protect their ship by deploying countermeasures such as decoys. However, figuring out which to deploy — should they send an interceptor to take the missile down or use radio frequency signals to confuse it? — can be challenging, as operators also need to consider the costs, timing, and laws related to different responses. As the size of a battle increases and new variables are introduced, the task of deciding which countermeasures to deploy, and when to deploy them, becomes more difficult.
Researchers in the Air, Missile, and Maritime Defense Technology Division at Lincoln Laboratory developed the Human-Machine Collaborative Optimization via Apprenticeship Scheduling (COVAS) system. COVAS uses machine learning algorithms to provide users with scheduling solutions that help them make decisions about the completion of specific tasks. These solutions mirror the ones that humans would make if they had time to consider all of the available options. COVAS is the first and only algorithm to provide real-time ship defense solutions by learning from human experts.
Developing efficient scheduling software to help users decide how to complete technical tasks, such as countermeasure deployment, has been a difficult task for the U.S. Navy and other organizations. Before COVAS, organizations would bring in a software developer to extensively interview industry-specific experts to understand the relevant policies and standards that an end-user must take into consideration when completing a task. This process is time-consuming and often does not produce software that mirrors human decision-making.
“Engineers and researchers alike fail to develop algorithms that actually meet the end user’s needs because they try to replace the user’s natural decision-making process with a synthetic one and because users cannot adequately describe their own natural decision-making process,” said Sung-Hyun Son, leader of the Ballistic Missile Defense System Integration Group, who helped Gombolay apply COVAS to a U.S. Navy resource-allocation project.
COVAS learns human operational behavior in real time by observing data input that comes from training schedules. These schedules detail specific actions that are taken (in the case of a naval battle, when and where a countermeasure is deployed) and actions that are not taken. COVAS only compares two actions at a time: the action a user took and the action a user did not take. These comparisons, known as pairwise comparisons, help COVAS predict which actions a user would take in a given situation and rationalize how resources will be best deployed.
“By only considering pairs of actions at a time, rather than all actions at once, COVAS is not limited to learning actions in environments with the same number of available actions,” Son said. “For example, I can choose an umbrella, jacket, poncho, or sweater if it rains. Say I choose umbrella given it rains. From that one decision, we can infer umbrella over poncho, umbrella over jacket, umbrella over sweater, no poncho over umbrella, no jacket over umbrella, and no sweater over umbrella. From that one decision, we got six training examples instead of a single datum with classical approaches.”
After learning from demonstrations of small problems, COVAS applies the sets of rules it developed from witnessing those demonstrations to more complex problems that have a greater number of variables to be considered. This capability allows the system to respond to both small- and large-scale problems without needing to observe additional behavior.
COVAS then provides users with a suggested schedule of actions that will help them solve their current problem. The real-time COVAS solution is based on a branch-and-bound approach. With this approach, the users receive a standard assessment score of how optimal the solution is, thus allowing the users to form their own opinions of the solution’s reliability before deciding on their course of action. The branch-and-bound approach additionally presents COVAS’s solution in a decision tree model. This model illustrates a course of action based on the statistical probability that the selected action mirrors the human decision-making process so that users can easily interpret how the system came to its conclusion.
References and Resources also include: