Helsing to Scale AI Underwater Surveillance

Helsing and Partners Scale Up AI Surveillance

Helsing and partners are preparing to increase production of a new AI-enhanced underwater surveillance package. Plans indicate early interest and the capacity to support trials or initial orders. Scaling will begin, with trials and testing underway. The partners designed the package to be more affordable than traditional methods of conducting underwater tasks like anti-submarine warfare (ASW), infrastructure security, or wide-area maritime surveillance. Lessons from the Russo-Ukraine conflict are being applied, including delivering more scalable, affordable systems.

Manufacturing and Deployment

Amelia Gould, Helsing’s General Manager Maritime, said at the Combined Naval Event (CNE) conference on 21 May that they plan to manufacture SG-1 in the UK, initially in the hundreds and eventually in the thousands. She added that the threat is real and immediate. The Lura/Fathom package focuses on building mass quickly. Gould noted that finding targets is increasingly challenging for humans, coupled with increasing threats to infrastructure and freedom of navigation.

Fathom/Lura Capabilities

The Fathom/Lura package provides sensing capability, allowing higher-end platforms like ASW frigates to conduct fixes. Gould noted that Western navies’ demands stretch across the Baltic Sea, Atlantic, and Pacific oceans, requiring a design that can operate worldwide. Fathom units can be operated as a contracted service or by navies directly. Helsing, in partnership with Blue Ocean Marine Tech Systems, Ocean Infinity, and QinetiQ, unveiled the package in early May. Helsing plans for the package to be available for deployment within the year, encompassing prospective procurement timeframes.

Gould said they are ready to deploy to sea and are establishing a UK industrial base to manufacture Fathom and support the capability at scale. She added that the Russo-Ukraine conflict highlights the importance of a resilient, simple, and scalable manufacturing solution. The capability has been in development for around 18 months. Initial AI benchmarking trials were conducted with the UK Royal Navy (RN), followed by further activities with two other navies. In-water trials have been conducted off the UK and Australia. Partnering with naval customers has shaped the package’s development.

The underwater battlespace is complex, making sub-surface platform detection challenging. NATO adversaries are operating more underwater platforms, while the alliance has limited coverage, as current ASW systems are complex and costly. Gould said that a new response is needed, requiring new thinking, solutions, and technology. She stated their vision is to search, deter, and defend across the world’s oceans with AI and autonomy at an affordable scale, becoming another feed into the maritime picture.

Gould added that the war in Ukraine has highlighted the need for numerous sensors and simple-to-produce, affordable, portable systems to get as many sensors in the water as possible. The package’s hardware and software are designed to be vertically integrated, offering enhanced AI sensing and processing on mass-deployable gliders. Fathom, a 2m, 60 kg glider, can be rail-launched from shore or at sea, including autonomously. It can be deployed from torpedo tubes or extra-large autonomous underwater vehicles. Traveling at 2-3 kt, Fathom uses buoyancy to glide, carrying passive sonar for sensing, with endurance reaching three months.

Originally designed for oil-and-gas sector environmental monitoring, Fathom has been modified for defense. Gould said they have adapted it for ASW to create a constellation of underwater sensors. Fathom uses Lura’s AI-enhanced onboard processing software, trained on historical catalogues of underwater acoustic data. Gould said that different data types for different regions are important to maximize Lura’s applicability across ASW use cases. The combination aims to improve coverage, volume, and speed in detection, analysis, and classification. Gould said that unlike a human operator it can do it 24/7, and one user can operate hundreds of gliders from a maritime headquarters.

AI and Autonomy

Massed sensing increases detection probability. Lura’s onboard processing performs sonar activities when detecting, classifying, localising, and tracking, giving navies a range and decision advantage. In April trials, Helsing ran simulation modelling surveillance across a 4,000 km2 area. Gould said that a single point of presence generated a 40 percent chance of intercepting an adversary, while a constellation of autonomous platforms achieved a 97 percent intercept probability.

Helsing’s concept of operations (CONOPS) involves deploying gliders as a massed constellation, scalable from securing a harbor to providing an ASW barrier. The initial CONOPS is that the gliders will not communicate, reducing hardware and software complexity while maintaining the deterrent effect. The gliders surface to transmit data, with onboard processing decreasing data volume and restricting the exposure window. Sensing and classification happen in real-time, with data transfer in near-real-time. The deterrence-based approach supports other CONOPS, like NATO’s ‘Digital Ocean’ vision. Gould said that massed gliders operating autonomously and reporting in near-real-time is what they consider digitising the ocean to look like.

Other lessons from Ukraine have shaped Helsing’s CONOPS. Ned Baker, Helsing UK’s Managing Director, said that to overcome communications degradation, they are putting AI on the edge, processing onboard the deployed vehicle. Further learning includes systems adapting from missions. Gould explained that Lura will learn from every mission, evolving at the pace of the threat. Baker added that they develop capability alongside end users, noting that this is the new paradigm of deploying defense capability. Working with stakeholders may generate capability evolution ideas, enabled by the software-defined capability package. A final lesson is Western navies’ need to trust in uncrewed technologies. The Fathom/Lura package’s software-based design and ease of operation will help build user trust. Gould said that the more time it spends at sea, the smarter it gets, and the more confidence you build in it.