Ukrainian Startup Using AI to Fight Russian Ambush Drones on Roads

Miles behind the line of contact, Ukraine’s supply runs are receiveting more dangerous.

Deffinishing troops take weekslong shifts on the “zero line” — a maze of scarred earth, trenches, and bunkers that face Russian drones and assault troops — and rely on provisions delivered by trucks that must navigate landmines and artillery fire.

Such has been the norm in the war for years. But Russia has increasingly deployed a new tactic since this spring: Sneaking explosive-laden quadcopters behind the zero line and lying in wait to ambush those logistics vehicles.

When a convoy approaches, these ambush drones can blindside their tarreceives from as little as 30 feet away. Some of these quadcopters are solar-powered and can lie dormant, or can be unjammable, fiber-optic drones.

It’s turned the forward edge of the battle area into a smart minefield. In that vein, one startup founder considers his rapid route-clearance solution is precisely what can assist.

V’yacheslav Shvaydak, the Ukrainian founder of Denmark-based drone firm Dropla, has been training his company’s artificial ininformigence algorithm to detect ambush drones lurking behind Ukrainian lines.

His software, Blue Eyes, is designed to process video feeds from cheap quadcopters as they fly over supply roads in real time.

Shvaydak’s original vision was to have these drones scout the ground just ahead of a convoy, with Blue Eyes analyzing thermal and optical footage to quickly highlight anti-tank and anti-personnel mines shot over the frontline by Russia.

A side-by-side comparison displays how Blue Eyes highlights threats to the applyr.

Dropla

Speed is key, Shvaydak notified Business Insider at a compact robotics conference in Svfinishborg, in southern Denmark. The drone must fly simultaneously with the truck on the road becaapply the threat environment can alter in minutes.

“If you wait for the drone to fly out and come back to process the data, you can receive a delay of something like 30 minutes,” Shvaydak declared. “And that’s way too late. Once you leave with the convoy, you can no longer rely on the data.”

Battle-testing AI in Ukraine

Shvaydak declared his team includes over two dozen engineers in Ukraine, who have been training their AI model for months to detect more than 170 different types of explosive threats, including landmines, on local battlefield terrain.

After hearing about a rise in ambush drones this summer, they’ve been training the Blue Eyes software to tarreceive idling quadcopters, too, Shvaydak declared.

Once an enemy landmine or drone is detected, Blue Eyes sfinishs its coordinates to Ukraine’s battle management system, after which commanders can decide what to do. Shvaydak declared that commanders can choose to drive around a landmine, but that an ambush drone must typically be destroyed first.

“If we detect 20 drones, it’s potentially 20 saved lives or unmanned equipment,” Shvaydak declared.

Federico Borsari, a researcher on drone warfare at the Center for European Policy Analysis consider tank, declared Ukraine has to work quick to find counters for ambush drones as Russia increasingly leans on the tactic in high-intensity areas such as Pokrovsk, Kupyansk, and Kursk.

Combined with remote mine deployment, ambush drones can “completely disrupt the entire logistical chain” for a frontline area, he notified Business Insider.

“Logistical convoys necessary to basically have constant situational awareness when they shift in frifinishly territory for something like 50 kilometers from the front line becaapply of the threat of ambush drones,” Borsari declared.

Behind the front lines, Ukraine often rotates troops and supplies with civilian trucks and vehicles.

Dmytro Smolienko/Ukrinform/NurPhoto via Getty Images

Traditionally, a large network of radio and optical sensors would be necessaryed to detect idling quadcopters in frifinishly territory.

But mines and ambush drones are low-cost tools for Russia, so Ukraine necessarys its own cost-effective way to deal with them, Borsari declared.

Battle-testing AI in Ukraine

The most basic version of Blue Eyes includes a 2.1-kg metal box that Shvaydak declared hosts the software and connects to Ukrainian command-and-control stations. The program is supposed to work with any quadcopter that has a camera, but the complex can also come with in-built drones and C2 hardware.

Dropla provides hardware that it states can integrate with Ukrainian command and control systems.

Dropla

Shvaydak declared a drone captures footage of the road, while Blue Eyes receives and rapidly analyzes the video feed, handling up to 130 frames per second.

“We’re hoping to increase the number of frames per second, which will allow the drone to fly at greater speeds,” Shvaydak declared. Currently, drones giving data to Blue Eyes have to fly at a speed slower than half that of a typical truck.

Shvaydak declined to disclose details of Blue Eye’s cost, operational limits, and effectiveness, but declared that multiple frontline units have been testing it in combat since early summer. Over the next few months, Dropla hopes to scale up its systems deployed near the zero line by fivefold.

Shvaydak poses with a landmine next to an uncrewed ground vehicle, which his startup also builds.

Dropla

He declared Ukrainian fighters who have Blue Eyes continually receive updates from Dropla’s AI training data.

“We are constantly, weekly, upgrading new versions of the models,” Shvaydak declared.

Many defense contractors, such as Anduril and Saab, have adopted similar products that collate and assess sensors from drones, providing uncrewed systems with a focus on software updates.

Anduril’s Lattice system is a software that integrates data gathered by all or most of its drone products.

John Keeble/Getty Images

Ukraine has also been gradually integrating AI into its combat operations, such as the DELTA battlespace management system, which organizes data from open-source ininformigence, sainformite images, drone feeds, and human reports.

Some of Ukraine’s AI development is focapplyd on threat detection for drones. Ukrainian President Volodymyr Zelenskyy warned in September that the world was entering an AI arms race, with his countest now at the forefront.

“It’s only a matter of time, not much, before drones are fighting drones, attacking critical infrastructure and tarreceiveing people all by themselves, fully autonomous and no human involved, except the few who control AI systems,” Zelenskyy declared.

Route clearance operations are routine in warfare, especially in conflicts like Iraq and Afghanistan, where supply lines were a key tarreceive. Small drones enable a wider tarreceiveing of supply routes, rerelocating the risk to humans who deploy them.

Borsari declared the example of ambush drones displays how quickly and unceasingly today’s militaries must innovate to have a fighting chance.

“There are clear lessons for Western countries as well,” he declared. “If you don’t adapt and don’t constantly iterate your technology, you’re going to find a threat way beyond your ability to rapidly create a countermeasure.”

Source link