The Coast Guard’s operational use of unmanned systems, begun in the air a few years ago, is accelerating. Six of its National Security Cutters (NSCs) – a number that will include all eight operational NSCs by spring of 2021 – are outfitted with unmanned-fixed wing aircraft that can be launched from their decks to expand maritime domain awareness (MDA) and provide actionable intelligence on hazards and threats. In 2018, the service’s Office of Aviation Forces began a pilot project to evaluate short-range aerial systems – remote-controlled rotary-wing vehicles – to assist with routine survey and inspection work.
The service did not get to the point of employing unmanned aircraft systems (UAS) operationally overnight. The Coast Guard’s Research and Development Center (RDC) began evaluating UAS nearly two decades ago, starting in 2001 with testing of advanced catapult launch and recovery technologies onboard CGC Harriet Lane.
Beginning in October 2020, the service took a decisive step toward introducing unmanned watercraft to its surface fleet. A 30-day pilot study was devised in response to a congressional mandate, issued in 2018, for the Coast Guard to evaluate low-cost, commercially available technologies that could enhance MDA in remote Pacific regions. The Coast Guard continues to face challenges with respect to conducting maritime surveillance necessary to support its statutory missions related to marine safety, security, and protection in the Pacific Ocean.
According to Cmdr. Blair Sweigart, PhD, chief of the Modeling, Simulation and Analysis Branch at the RDC and Pilot Demonstration Director, Congress didn’t specify the platform to be used in the pilot; the Coast Guard’s choice of systems was guided by the focus on IUU fishing. “It’s a global issue,” said Sweigart, “and we stand alongside a lot of partner nations in trying to combat this threat. The choice to explore unmanned surface vessels was driven by the desire to keep the costs low and accessible for our partner agencies and potentially partner nations.”
The Coast Guard’s request for proposal included several requirements: the unmanned surface vessels (USVs) to be evaluated must be able to operate for 30 consecutive days without refueling; to detect other vessels within a minimum of one nautical mile; to provide notification of this detection within 6 hours; to operate within a defined area of 20 square nautical miles off the coast of Honolulu, Hawaii; to operate solely by beyond-line-of-sight communications; and other requirements. In February of 2020, the service awarded two contracts for operation and support of USVs during the 30-day pilot: to Saildrone, a manufacturer of wind- and solar-powered USVs; and to Spatial Integrated Systems, Inc. (SIS), a developer of robotic control systems.
The evaluations were conducted as part of the Low Cost Maritime Domain Awareness Pilot project under the RDC Surface Branch and began on Oct. 7, off the south shore of Oahu, as Sweigart and four colleagues from the RDC ran the USVs through a series of simulated exercises. Five other members, including Project Manager Scot Tripp, remained at the RDC in New London, Connecticut, to perform remote monitoring.
The Saildrone is a remotely operated surface vehicle, about 23 feet long, propelled by wind acting on a 15-foot tall rigid sail. Its sensors and other equipment are powered by solar cells. Because they are powered by sunlight and wind, Saildrones can operate for months at a time, traveling programmed routes that can be modified by remote operators. They have an established record of conducting scientific missions in remote and often rough environments: At least one Saildrone has circumnavigated Antarctica, and from May to August, 2020, four Saildrones, at speeds of 2-3 knots, sailed from company headquarters in Alameda, California, to Point Hope, Alaska, to conduct hydrographic surveys off the North Slope for the National Oceanic and Atmospheric Administration (NOAA).
The six Saildrones used for the Coast Guard’s MDA pilot conducted simulated patrols over Penguin Bank, a coral reef 25 miles southeast of Honolulu. The drones were equipped with two mechanisms for identifying vessels: optical cameras for visual sightings, mounted atop each drone’s sail; and Automatic Identification System (AIS), the digital tracking system used by vessel traffic services. Saildrone cameras captured images every five seconds and completed onboard analysis of every image, using the company’s proprietary artificial intelligence (AI) software, and then sent confirmed vessel images to Saildrone command centers. All were controlled via an internet web portal that provided imagery and tracking data for each detection – and also provided a means of re-tasking the Saildrones as they conducted several MDA activities.
According to Sweigart, the Saildrones were positioned in a variety of formations throughout the demonstration to simulate patrol missions, including a picket-line formation to detect vessels moving from one area to another, and simulated closed-area monitoring formations to identify vessels moving into a restricted area. “We also looked at port security vignettes, positioning the Saildrones as if they were in a port,” he said, “and then looked at what contacts and what information we could glean about vessels that were coming into the formation of Saildrones. We also had the drones run different search patterns just to see what that might look like as we consider how we could employ this technology across our mission set.”
For the 30-day pilot, SIS put together a customized vehicle, “The Watcher,” based on the Coast Guard’s cutter boat-large platform, a 7-meter interceptor-style boat powered by an inboard diesel engine capable of 30 knots. To meet the pilot’s endurance requirements, the boat was equipped with an auxiliary fuel bladder. Onboard systems were powered by a battery bank, which was itself charged by a solar array panel.
Equipped with commercial radar and an electro-optical infrared (EO/IR) camera, The Watcher patrolled an area 15 miles southwest of Honolulu, remaining within its 20-square-mile boundary and providing alerts every hour when another vessel had been detected. The boat was operated primarily out of SIS headquarters in Virginia Beach, Virginia, via satellite, with support and additional communications provided out of a command-and-controltrailer at a Honolulu shipyard.
The Watcher’s radar detected other vessels up to five-miles away and its camera recorded images every 10seconds. The Watcher also conducted simulated patrol scenarios to demonstrate patrol and search pattern execution as well as target detections and investigation activities. The Saildrone has much greater endurance than The Watcher, which is powered by internal combustion, but the Saildrone is also significantly slower and less maneuverable; The Watcher, Sweigart said, “has a much more capable top speed in terms of investigating a contactand getting additional imagery or sensor data if it detects something.”
29RDC: COAST GUARD-OWNED AND OPERATED
The Saildrone and SIS systems evaluated during the30-day pilot were contractor-owned, contractor-operated USVs – which is one way of doing things, and can be the most cost-effective alternative when dealing with complex emerging technologies. But the demonstration included a third USV, purchased by the RDC to evaluate the efficiency of government-owned and operated systems: 29RDC, a 29-foot autonomous boat built by Metal Shark of Jeanerette, Louisiana, with robotics and control systems supplied by Sea Machines of Boston.
29RDC’s autonomy suite is a self-driving system that allows the vessel’s wheel and throttle to remain functional while still allowing manual operation if a crew member is aboard and wants to take control. “It could be run without a crew,” said Sweigart – but crew members were aboard at all times during the demonstrations for safety. “It opens up the interesting possibility of a hybrid model where the vessel can drive itself, even with the crew on board, and you can free the crew up to do other things and keep their attention focused outside of the vessel, as opposed to on the inner workings of the boat,” Sweigart said. “We looked at this one a lot in terms of a search-and-rescue platform, but that’s not to say it’s any less capable of performing some of those other mission sets.”
The Coast Guard’s experimental USV accomplished dozens of missions during the pilot including navigation, search patterns, collision avoidance, and remote control. 29RDC was operated multiple times by the RDC team in New London, 5,000 miles away, using cellular service.
FINDING THE EDGE
Shortly after the pilot’s operations ended on Nov. 5, 2020, the RDC team composed a brief “quick look report” offering an overview of operations and some preliminary conclusions. A more detailed report will follow in 2021.
The demonstrations revealed strengths and shortcomings, both for different kinds of systems and for the service-provider and government-owned model of deployment. Commercially available USVs are clearly capable of performing some level of daytime MDA missions, and can operate independently for up to 30 days. But each system evaluated in the pilot has some limitations, Sweigart said – which were meant to be discovered as his team designed the exercises. “Both companies were willing to work with us to push the boundaries and see what the assets were and weren’t capable of,” he said. “I thought that was hugely important … Within the first couple of days, we can see what the assets can do. We need to then push the envelope to see: What is the edge? Where do we start to maybe lose that ability?”
Given the strengths of the different systems, the team foresees a “layered” use of USVs in the Coast Guard’s future, one that combines the endurance and autonomy of USVs such as Saildrones with The Watcher’s ability to pursue and collect more information. Future solutions may also combine the advantages of the service-provider model with other modes of acquisitions, such as government-owned/contractor-operated, and government-owned/ government-operated – depending on how the asset is used. As these automated pieces fall into place, the Coast Guard may determine the need for a manned asset, or a human watch stander aboard an autonomous asset, to be added to a mission.
Another key lesson from the pilot is that artificial intelligence and machine learning may well prove indispensable to future deployments of USV technology. While all three systems collected a wealth of MDA data, only one was capable of conducting on-board processing of this data – speeding up processing and limiting bandwidth consumption by limiting the transmitted data to actionable images.
As the Coast Guard continues to explore USV technologies, said Sweigart, it will look for this capability, and also the potential for a system to identify the additional data that might be useful to provide Coast Guard watch standers – or even the ability to act on data itself, approaching a target of interest to try to get better data.
“I think there’s a much longer-term potential for an overarching AI system to help understand the data that’s coming back and additional action that may be needed,” Sweigart said. “And if an additional asset is needed, maybe that AI system can provide that data in a more actionable format to the Coast Guard watch stander, and prompt them to say, ‘Hey, here’s what’s going on,’ and make the best use of the humans in the loop.”
One of the most important things the 30-day pilot has revealed to the Coast Guard that its intense interest in USV technology is shared by a multitude of government partners. As news of the upcoming evaluations circulated, several federal partners engaged; the pilot partnership eventually included the participation of U.S. Customs and Border Protection, CBP Air and Marine Operations, the Navy Research Laboratory, the Office of Naval Research, NOAA, Scientific Advisors for the U.S. Navy, The President’s Intelligence Advisory Board, and Congress. This strong coalition is certain to spur further investigations and interest in USVs.
“I think we understand the edge for the capabilities that we examined in the demonstration,” Sweigart said. “But the technology is continually evolving. There’s certainly room for continual assessments, and hopefully we can start to make those assessments through employment. There’s a lot of interest in looking at how we can transition this into an operational reality to augment Coast Guard missions.”