Fighting suspicious objects: how automated technologies will increase the capabilities of Russian UAVs

Rostec specialists have created a self-learning program capable of detecting suspicious objects with high efficiency.

RT was told about this in the press service of the state corporation.

The developer of the program is the Central Research Institute "Cyclone" (Moscow) of the holding "Ruselectronics" (part of "Rostec"). 

“The principle of operation of a self-learning program is based on memorizing suspicious objects and then comparing their images with new targets.

Thus, objects that pose a potential threat are identified, ”the state corporation said.

As noted in Rostec, the program is able to accompany low-contrast and small-sized objects in any range of the spectrum.

At the same time, its algorithm is resistant to sudden camera movements, changes in the angle and scale of the image.

Rostec is confident that the brainchild of the Central Research Institute Cyclone will find wide practical application, including as part of small-sized equipment carried by UAVs.

“The algorithm can be used in infrared and television surveillance systems.

It can also be used as part of UAV optical systems to detect mines.

In addition, the program can interact with a neural network that performs primary target detection for subsequent tracking (tracking. -

),” the press service of the state corporation explained.

The object for tracking can be set by the operator by highlighting its image on the monitor screen.

When a suspicious object is first detected, the tracker starts tracking it using video from a drone.

As explained in Rostec, such a scheme allows "to verify the target from various angles for the final confirmation of its danger by the neural network."

“The program does not require significant computing power.

One standard processor used in portable devices is enough to track up to four objects in real time at a change rate of at least 25 frames per second, ”Rostec emphasized.

"Rapid Implementation"

In an interview with RT, the editor-in-chief of the Unmanned Aviation publication Denis Fedutinov suggested that the algorithm created by the Cyclone Central Research Institute would solve such an important problem in the UAV segment as automating the processes of detecting, identifying and tracking enemy objects, including enemy equipment and manpower .

“This will significantly facilitate the work of operators who need to visually process a significant amount of data from specific reconnaissance,” Fedutinov said.

• UAV of the RF Armed Forces

• RIA News

• © Konstantin Mikhalchevsky

A similar point of view in a RT commentary was expressed by the director of the Center for Unmanned Aviation, adviser to the Russian Academy of Engineering Maxim Kondratiev.

“The use of these technologies will significantly reduce the load on the UAV calculation, since the process of searching, identifying and classifying objects occurs in a fully automatic mode.

As a result, the fatigue of military personnel is reduced, and the productivity and quality of control increase dramatically, ”the expert said.

According to Kondratiev, automation of the processes of working with targets will have another positive effect: it will certainly speed up the training of drone operators, which is especially important in a special operation.

The expert recalled that now the army is undergoing mass training in the management of UAVs, including mobilized ones.

Identification of targets is one of the key stages of their preparation, since in the theater of operations (theater of operations) the enemy often resorts to camouflage.

“It must be said frankly: there is neither time nor resources to make a high-class UAV operator out of every serviceman.

The servicemen are given a base, and the nuances of work have to be mastered already in a combat situation.

However, the program announced by Rostec should become a good assistant to our servicemen at any time of the day in the combat zone,” Kondratyev said.

At the same time, according to the expert, the equipment that ensures the operation of the algorithm is unlikely to be expensive.

Thus, the Russian industry, as the interlocutor of RT expects, can deploy its mass production for the speedy deliveries to the part of the RF Armed Forces involved in the special operation.

“As far as I understand, significant computing power is not required to ensure the operation of the system, and therefore the cost of the technology will be very moderate.

This will help a wide and rapid introduction into production,” says Kondratiev.

The expert called the technical possibility of fixing explosive devices embedded in the algorithm a significant achievement of the Cyclone engineers.

According to Kondratiev, the theater often has to carry out engineering reconnaissance of the area in order to detect minefields and obstacles, unexploded artillery shells, caches of enemy ammunition.

“Before the sapper groups enter the area, a drone with this program will collect the most complete information and plot dangerous zones and objects on an electronic map.

This will reduce the demining time and provide additional safety for the work of specialists, ”the expert emphasized.

According to Kondratiev, at present, UAVs with a self-learning program or a neural network that allows detecting mines would be in demand in the NVO zone and, in particular, in Donetsk, over which, with the help of special MLRS shells, Ukrainian troops constantly scatter anti-personnel mines "Lepestok", prohibited international law.

• Preparation of UAV "Orlan-10" for launch from a catapult

• RIA News

According to Fedutinov, the technical possibility of detecting explosive devices from drones is ensured by the use of optoelectronic sensors.

They allow you to find mines with a high probability both in the ground and on its surface.

“The search for explosive devices from the air is carried out by fixing characteristic unmasking signs, both direct and indirect, for example, violations of the surface of the soil and vegetation, temperature anomalies, changes in the level of soil moisture, and the like,” Fedutinov explained.

Everything in the complex

According to experts, the physical carrier of a self-learning program or a neural network on a UAV, as a rule, is a hardware-software complex (HSC).

One of these products was developed in 2020 by specialists from the Era military innovative technopolis together with the Department of Information Systems of the Russian Defense Ministry and Russian defense enterprises.

The main purpose of the product is the maximum automation of reconnaissance and strike assets of the Russian army.

The PAK integrates an interspecific data bank of "reference portraits of weapons, military and special equipment."

On their basis, the program is trained (in this case, the neural network).

On the theater of operations, the complex allows you to track the movement of enemy troops, even if the enemy uses various methods of camouflage and tries to take advantage of the dark time of the day.

“In particular, the complex can be installed on the drone of an army reconnaissance unit.

Thanks to our product, the UAV gets the opportunity to identify and classify the observed objects of the enemy, ”Major Yevgeny Nazarov, deputy head of the research department (search and predictive research) of the technopolis, told RT.

As Denis Fedutinov explained, PAK is a product of the development of artificial intelligence technologies.

According to the expert, at present, without their use, it is impossible to effectively and quickly process an array of information about the situation in the theater.

“With the use of a software package based on artificial intelligence, the military is able to process data received from several surveillance channels.

If all processes are well debugged, then the effectiveness of aerial reconnaissance is seriously increased, ”Fedutinov concluded.