Ukraine's drone resurgence: AI technology shifts battlefield dynamics
The war in Ukraine is the first conflict where drones have been used on such a massive scale by both sides. Initially, Ukrainian drones were tremendously effective, but their efficiency drastically decreased due to the widespread use of jammers by the Russians. However, for a few months now, it appears superiority is returning, owing to the implementation of artificial intelligence in FPV drones.
The Ukrainians are currently deploying hundreds of thousands of drones for various tasks, but the most famous recordings are of FPV drone attacks. These drones are used to attack not only equipment such as BMP-3 infantry fighting vehicles and tanks, including the latest T-90M, but even individual soldiers and helicopters.
After the Ukrainian drones' initial impressive effectiveness, the Russians began to extensively deploy jammers to disrupt control or GPS signals. This resulted in a classic 'shield and sword' race, with periods of drone dominance or impotence during which experiments with control signal frequencies were conducted.
As with professional military drones, the solution to these problems was sought in artificial intelligence algorithms enabling autonomous attacks on designated targets. After months of trials, the Ukrainians, sometimes with Western companies' assistance, achieved satisfactory results. Below, you can see a crashed FPV drone supported by a Google Coral AI module.
FPV drones with artificial intelligence — only physical elimination ensures protection
The key is to provide a module with sufficient computing power to create computer vision capable of tracking and following a selected object. Due to the limitations of vehicle-mounted jammers, their effective range is about 400-500 metres, and fully autonomous flight must be adequate to cover such a distance.
Today, there are many commercial AI solutions that are quite capable of object recognition. The difference is that, instead of humans or, for example, kittens, the algorithm must recognise, for instance, a T-72 family tank and follow it. Some solutions, such as Skynode-S modules, even provide navigation by triangulating the drone's position and comparing the terrain seen by the drone’s camera with preloaded satellite maps.
One challenge is developing machine learning algorithms, and another is creating systems with sufficient performance, compact dimensions, and low power consumption. As a result, a drone is developed that can only be stopped by shooting it down.
In professional solutions, these include the Israeli SMASH modules from Smart Shooter, which have gained significant interest in recent years among the armed forces and enable the shooting down of a drone with handheld firearms even from several hundred metres away (effective range varies with the calibre of the weapon it's mounted on).
In the case of improvised solutions, 12-gauge smoothbore shotguns perform well. These, with appropriate buckshot ammunition, can be effective even at distances of about 50 metres, and with the use of new ALDA ammunition from Beretta Holding, it is possible to target drones at distances of approximately 80-120 metres.
Google Coral AI development board — the perfect solution for prototyping
The photos show that part of the drone includes a G950-04742-01 development board, which is characterised by a performance of 4 trillion floating-point operations per second (TOPS) while requiring only 2 watts. For example, Google asserts that this system can execute the MobileNet v2 SSD neural network architecture instruction in 14 milliseconds, while a 3 GHz Intel Xeon Gold 6154 processor requires 106 milliseconds.
The Google board features a ready-made NXP's iMX 8M system with four ARM Cortex-A53 cores, an ARM Cortex-M4 core, integrated Vivante GC7000Lite graphics, Google’s proprietary Edge TPU coprocessor, and a cryptographic coprocessor. Additionally, it has 8GB eMMC memory, 1GB LPDDR4 memory, and necessary I/O ports.
Such boards are available for purchase without restriction, and the Ukrainians, unlike the Russians, have unhindered access to them. However, as the sanctions reality shows, a certain number are likely to be available in Russia.
