The developers have improved the drone detector, which is widely used in the area of its
A new version of the Bulat drone detector is being prepared for mass production, Izvestia found out. It has integrated artificial intelligence, which has been trained to scan more than 150 unique frequency signatures of various UAVs, said the developers of Bulat, a team of 3mx volunteers. The innovation makes it possible to increase the sensitivity of the equipment and at the same time reduce the number of false positives. The use of artificial intelligence in drone detectors is a promising direction, since it copes better than the operator with the allocation of a weak electromagnetic signal in the general flow, especially against the background of interference, experts say. About the new development of the "national" military—industrial complex - in the material of Izvestia.
The introduction of AI into military devices
The Bulat drone detector of the fourth version with artificial intelligence will be put into mass production by the end of May. Sergey Shandobylo, Director of development of the 3mx team, told Izvestia about this.
The built—in AI can distinguish UAV signals from wireless networks - WiFi routers, Bluetooth devices, cellular communication stations - even if their signals overlap each other in a certain frequency band. It is also capable of detecting drones with low-power transmitters, whose signals are difficult to distinguish against the background of interference in the air.
Drone detector "Bulat", 4th version
Image source: Photo: 3mx Team Press Service
— First, we manually made several copies, conditionally, soldered all the components on a homemade board on the knee, — he noted. — Then we checked whether the device was working or not. Tests were carried out, and according to their results, the Ministry of Defense recommended Bulat v.4 for purchase. That's why we're launching it into a series. And now the components will be applied to the board on the factory machine.
The tests took place both in the measuring laboratories of the 3mx team and at the landfills.
— The artificial intelligence developed by our team has already been trained to scan more than 150 unique signatures of various types of drones, — Sergey Shandobylo emphasized.
The fourth version of the detector covers a wide frequency range: from 300 to 6200 MHz.
Additional amenities
In addition to the Bulat, the 3mx team has created an individual notification device "Tair", which will also be released in the series at the end of May. It is made in the form of a wrist smartwatch, automatically connects wirelessly via Bluetooth Low Energy and operates at a distance of up to 30 m from the detector.
When a drone is detected, Bulat v.4 sends all data to the Tair display. As a result, the type or brand of the drone, its frequency and the time of the last detection are displayed on the watch screen. "Tair" can display information from all "Bulat" v.4 or only from a linked detector.
"Bulat", 4th version
Image source: Photo: 3mx Team Press Service
The device also notifies you when a drone is detected with a vibration signal.
Advantages of artificial intelligence
Now there is a very serious technological competition between Russia and the collective West, which includes the development of military products using AI, Alexander Rodin, executive director of the Scientific and Technical Center for Environmental Monitoring and Ecology at MIPT, Doctor of Physical and Mathematical Sciences, told Izvestia.
"Artificial intelligence is not a panacea for FPV drones, but it is a necessary technology," he explained. — We have already learned how to implement it on fairly simple and cheap electronics, so there are simply no obstacles to scaling and mass distribution today. However, it is precisely in the creation of drone detectors with AI that the enemy has more experience. The APU uses a mobile phone, puts it in a birdhouse to protect it from the rain, and hangs it on a tree. Now the whole territory of Ukraine is studded with them. As soon as their drone detector hears the sound of our "Geranium", it transmits it further into a single system that determines the trajectory.
Photo: IZVESTIA/Dmitry Korotaev
Image source: iz.ru
At the same time, according to Alexander Rodin, our engineers are ahead of the enemy in using AI in FPV drones.
The peculiarity of artificial intelligence is in its self—learning. That is, the situation is changing, new nuances in the signals appear, and he learns to distinguish these things from the general information noise, explained military expert Dmitry Kornev.
"After seeing something incomprehensible several times, the AI, with the help of an operator who says, "Yes, this is a problem," can gradually learn to understand that this kind of anomaly is a new drone," he clarified. — This is important because manual work of this kind is labor-intensive. But it is impossible to teach this, for example, computers of the 1980s, 1990s, and 2000s, because they need to clearly describe the rules. And artificial intelligence needs to describe a certain general algorithm, and then, accumulating experience, it begins to learn by itself.
Previous versions
Bulat was established in 2023. Two months passed from the idea to the release of the first serial sample — one month was under development, another one was put into serial production by the team. And all versions were in demand on the front line, Sergei Shandobylo stressed.
— As soon as we released the third version of Bulat, we discontinued the second one. We are currently producing an additional module for the third version. It helps the device to increase the range of detected drones," he explained.
Photo: RIA Novosti/Stanislav Krasilnikov
Image source: iz.ru
When a UAV is detected, the device notifies the user with vibration and a light indication. The module transmits information to all Bulat devices that are within 30 m of it.
The drone detector operates in passive mode, scanning the radio frequency spectrum 360 degrees, without radiation into the ether. It reacts only to UAVs, excluding false alarms on other signal sources. Detects them at a distance of up to 1.5 km.
Hacked, studied, created
The 3mx team includes engineers with many years of experience in radio electronics, communications, development and production of devices and sensors of various complexity. Focusing on several areas, they are trying to meet the needs of our fighters in detecting and neutralizing UAVs.
— Since the beginning of our work, we have been analyzing the market, studying what is available at the moment, how these funds are coping. We found out that there are devices from our manufacturers, but they are often expensive and small-scale. Because of this, a simple fighter on the front line, who wants to protect himself by spending his own savings, cannot do this. Either there is not enough money, or they simply do not sell it because of supplies only to the Ministry of Defense," said the development director.
The soldier of the VFU
Image source: Photo: REUTERS/Aleksander Ratushniak
The 3mx team started buying foreign analogues, studying them, opening them, hacking. And already on the basis of the knowledge gained, they started manufacturing their own equipment. According to Sergei Shandobylo, initially anti-drone guns were made in industrial quantities, now several thousand Airat-100s are working at the forefront, then the Bulat drone detector was developed, the first version began to be delivered more than a year ago.
Julia Leonova
Denis Gritsenko