Blinding eyes in the sky

If the war in Ukraine has taught us anything, it is that drones can make or break a campaign. Providing vital intelligence on enemy positions and payloads to destroy enemy armour, UASs (unmanned aerial systems) have proved their worth in the Donbas and beyond. Just a week into the conflict, to give one example, a 15-year-old used his commercial drone to spot enemy vehicles before relaying the information to Ukrainian artillery units. By one estimate, the boy and his father helped destroy nearly two dozen Russian tanks by shortening the time it takes to perceive a threat and act appropriately.

Yet amid growing acceptance in the power of drones – the Kremlin alone has spent around $9bn on UAS in the last few years – the systems to repel them are often surprisingly weak. Leaning on outdated personnel-intensive platforms, sometimes just re-purposed anti-aircraft platforms, many traditional CUASs (counter unmanned aerial systems) feel inappropriate in a world where drones can dominate the battlefield. Not that the situation is hopeless. By leveraging the latest processing technology, and keeping up with broader developments in the field, militaries can protect themselves from any airborne threat.

Accelerated kill chains

Tristam Constant is in a good position to reflect on the opportunities and challenges of modern UAS. A soldier in the British army for a decade, he now works at Anduril, a defence products company building AI-enabled autonomous systems, focusing on CUAS growth and delivery.

And as Constant stresses, UAS has proved its worth time and again across the war in Eastern Europe. “Low cost UAS are being utilised on both sides of the current conflict to accelerate kill chains,” he explains. “I think the fact that they’ve been so prevalent demonstrates their value on the battlefield.”

As the anecdote of the tank-busting teenager implies, the main strength of contemporary UASs are their versatility. Costing as little as $1,000, users with some savings and access to a smartphone can use them for a range of intelligence tasks. That is shadowed by more direct lethality. 3D printers have helped arm some drones with canister bombs, while the Biden administration plans to sell Kyiv UASs that can be armed with Hellfire missiles. Yet study what militaries are doing to counter this enemy in the sky, and it is clear that CUAS is far less dynamic. That begins with manpower. Some contemporary platforms require as many as 20 people to operate effectively. This is bad enough at the best of times but even worse in a war like Ukraine’s, where casualty rates can reach 1,000 per day and where artillery can decimate a huddle of troops in an instant.

This is ultimately a question of technical capacity. Lacking so-called mesh networking and edge computing – avoiding the need to send sensor data back to a central hub, speeding up processing and reaction times – many existing CUAS systems require troops to congregate in a single spot. Apart from being more dangerous, this approach is also easier to disrupt. “For a point-to-point network,” Constant explains, “it's very easy to jam that sensor, comms link and control node.”

That is echoed by broader technological problems. Relying on general-purpose radar systems, for instance, risks exposing the device’s signature, leaving the platform vulnerable to enemy attack. Even worse, they struggle to accurately judge the elevation of UASs even as they can pinpoint their distance. Constant summarises the challenge succinctly: “Many suppliers who produce CUAS systems may not have factored for their use under current warfighting conditions.”

Nor are they necessarily the only ones to be stuck in the past. Constant, for his part, suggests that some Western militaries still consider UASs a mere “unconventional threat” rather than the conventional danger they now are. More to the point, this attitude has serious practical consequences. Quite aside from the columns of smouldering tanks, the war in Ukraine has proved that traditional CUAS systems force soldiers to stay in limited air defence bubbles, seriously hampering their manoeuvrability.

Robust technical skills

How to counter these difficulties? For Constant, the answer begins with treating CUAS not as something that can be solved quickly, but rather as a “problem set” that requires multiple overlapping solutions. From there, he argues, insiders need to be “agile enough” to keep up with a threat that is continuously evolving.

Certainly, Anduril’s work suggests the company is following both these principles to the letter. Based in California and funded by venture capital, the company can hire the best talent and keep up with UASs as they evolve at speed.

If nothing else, this is reflected in the practical work Constant and his team are doing. For starters, Anduril works to “fuse different sensor modalities together” – helping CUAS operators locate incoming aerial threats quickly and efficiently, unlike older radar-based platforms.

From there, the company uses AI to make its CUAS platforms less manpower intensive than traditional alternatives. Rather than having dozens of troops analysing sensor data manually, Constant instead explains that his company's software layer, Lattice, “is able to process and fuse huge quantities of sensor data at pace – by leveraging computer vision and machine learning – then escalate and prioritise threats to human operators meaning they can take actions and seamlessly task their effectors as fast as possible.”

Anduril is doing similar work in other areas. Exploiting secure mesh networking, for example, means that CUAS platforms are far less open to cyber attacks, while edge computing offers greater redundancy if something goes wrong. At the same time, integrating signature-free systems helps keep CUAS platforms safer from enemy intelligence.

Robust underpinnings

It goes without saying that this technical wizardry enjoys robust philosophical underpinnings. Rather than simply fulfilling a contract then moving on, Anduril instead offers what Constant describes as CUAS “as a service” – allowing the company to “build and adapt our products to keep pace with an ever-evolving threat picture.” This is underscored by Anduril being awarded a SOCOM Systems Integration Partner contract for the next decade, valued at $1bn. In large part, Constant and his team are able to be so flexible thanks to their company’s core focus. As these references to AI and mesh networking imply, Anduril is a software first provider, one that leverages what Constant calls an “open architecture philosophy” to seamlessly integrate new sensor types and modalities into CUAS without needing to constantly build new hardware. Given how quickly UAS technology is developing – the military drone market currently enjoys a CAGR of 6.6% – this strategy seems wise.

That is doubly true in the specifics, with Constant describing how UAS makers are always looking for ways to up their game, for instance by masking a particular radio frequency signature or mitigating the effectiveness of a specific sensor. Fortunately, Anduril can also keep abreast of these developments thanks to its close links with defence ministries. “We’re really focused on building the products that the defence industry needs and wants,” Constant explains. “We build collaborative partnerships with defence customers to really understand their requirements.”

The real-world benefits of these relationships are equally obvious too, for instance when Anduril encountered a software problem and was able to write code mitigating the vulnerability over the space of a single day. No wonder Constant says his company boasts a “very aggressive” software release cycle. All this bodes well for the future, where the cat-and-mouse fight between UAS and CUAS shows no sign of abating. “The field will continue to evolve at a rapid pace,” Constant argues, adding that both growing government enthusiasm and low costs will force CUAS experts to stay on their guard. A fair point – even if his team at Anduril is clearly well-placed for whatever comes their way.