Russia’s V2U autonomous drone is striking targets in Ukraine

Russia has been testing and deploying its V2U autonomous drone in Ukraine according to a series of posts on social media and a news release from Ukraine’s Main Intelligence Directorate on the 9th June.  The Main Intelligence Directorate (HUR) has released information on them based on captured systems, but Serhii Flash on Telegram has been reporting their presence since May, and there are some indications that they have been in use since February 2025. 

Serhii has also posted on the strikes conducted by the systems and the way that they are being used, with one post on the 16th May indicating that they were active in three areas of the front, with 30 – 50 sorties per day. His assumption at the time was that the Russians were working to train the artificial intelligence used by the drones to recognise and detect targets, which could be valid if they were being recovered or the information otherwise relayed back to help train other systems. 

On the 24th May, Serhii reported on an attempted strike by the V2U against targets in the village of Velykyi Burluk, where he said a group of seven drones formed a circle above the village before reportedly diving to attack a cluster of cars and people in the market place below. One of the drones had colourful markings on its wings, which enables the drones to recognise each other, he said. This followed an attack on the 21st, which employed a drone with a petrol engine – as opposed to the electric ones used by other platforms. 

The report published by Ukraine’s HUR notes that the drone is equipped with LiDAR, a high quality camera, GPS module, and radio relay, as well as a Chinese minicomputer called Leetop A203, which uses an NVIDIA Jetson at its core. The Jetson single board computers are used by many companies that are developing AI-enabled systems for defence applications, including Helsing and Skydio, amongst others. The Jetson computers are relatively small and cost-effective for the computing power that they provide, which enables them to be deployed on a space-constrained system like a drone and provide AI at the edge. 

The combination of navigation systems described above gives some indication of what enables the V2U to be autonomous. The camera and Jetson together can enable computer vision, which can be used for navigation with a set of pre-loaded images that are uploaded to the drone’s computer before flight. Jetson is able to run AI models on the drone and the data captured by the camera. “A feature of the drone is its ability to autonomously search and select targets using artificial intelligence,” the HUR article states. If the models have been effectively trained, they could autonomously detect, identify, and engage Ukrainian forces. There is likely a lot of nuance involved in this assessment, however. 

A recovered V2U autonomous strike drone. The parts are mostly sourced from China and western companies. Credit: HUR

For instance, detection and recognition of armoured vehicles and howitzers is very achievable, this is because the vehicles are so different from anything else that the drone is likely to detect. However, it is apparent that Ukrainian and Russian forces are relying upon civilian vehicles for a lot of their transport to and from the frontlines, which would be very difficult for the drones to target without seeing soldiers in uniform enter. That said, the HUR indicates that the V2U can be flown as an FPV, enabling the operator to manually engage a target if required. Control for the drone in that case is carried out using the LTE communication – the wireless broadband standard that is often used for mobile phones. It sends its traffic over a Ukrainian network, requiring the drone to carry a Ukrainian SIM card. 

Computer vision can be unreliable, especially if the terrain has been changed extensively by fighting. However, the V2U has a LiDAR sensor that is facing down, which indicates that it is either used to measure the distance between the drone and the ground, or may be used for navigation in a similar fashion to the Tactera system from Flarebright. Combined with the GPS, this likely means that the drone is quite resilient to jamming, with the ability to either fly autonomously to a given direction, to be piloted there, or to navigate using GPS.

The warhead weighs 3 kg, and the available footage indicates that it can conduct a near vertical dive onto a target, which would complicate air defence if the drone is not intercepted before the dive. Contrast this with Lancet, Russia’s primary loitering munition, which often has a more shallow dive towards a target, enabling interception or other action to avoid the strike. Nonetheless, Ukrainian forces have intercepted the V2U using drone interceptors, so it is far from infallible. The range using an electric motor is 80 km, although the petrol engine extends this to 100 km, according to Serhii. 

Calibre comment

How dangerous is the V2U, and what does it bring to the battlefield? 

The short answer is that its primary strength is in resistance to electronic warfare (EW), which is still responsible for constraining FPV and other drone usage and attriting small drones across much of the frontline. The ability of EW to do this has fluctuated throughout the war in the cat and mouse game of Ukraine, but generally speaking has remained an effective defence against drones of all types and even some guided munitions. If the V2U is capable of autonomous navigation, then it would eliminate the vulnerability to EW to a large extent, forcing Ukraine to rely on kinetic interception, in a similar fashion to fibre optic FPVs.

However, the constraint with fibre optic FPVs is that they are limited to one drone per operator, the V2U has demonstrated some swarming capability, which would likely conserve manpower or enable deployment at much greater scale during an offensive or defensive operation. The autonomous targeting element is concerning, especially if the models are being trained to engage civilian vehicles. It may eventually enable the detection and targeting of Ukraine’s own drone hides, which would be problematic, and could definitely engage conventional armoured vehicles. However, with those mostly kept away from the frontline, the opportunities to find and engage them might be limited. In addition, Ukraine is receiving its own stocks of AI-enabled strike drones like the earlier variant of the HX-2 from Helsing and a similar system from Stark Defense, so the Ukrainian Armed forces are not without their own capabilities in this arena. 

In sum, the autonomous navigation is perhaps the most interesting aspect, especially when combined with the ability of an operator to take the drone onto a target remotely. It is more difficult to assess the utility of the autonomous targeting at this stage, because it is not clear what the drones have been trained to engage. Moreover, there is genuinely an early adopter advantage with AI in that the developers of the V2U will be gathering operational data on which to train their algorithms and models, which could dramatically improve their capabilities over time. 

By Sam Cranny-Evans, published on 10th June, 2025.