The special operation emphasized the role of precision weapons
The United States and Britain have today retargeted their commercial satellite systems Starlink and OneWeb to collect intelligence information in the interests of the armed formations of Ukraine. A plan is being hatched to disconnect Russia from the GPS system. How can all this, coupled with the ban on the import of software (software), affect the performance of combat missions by our troops in the combat zone, the production of high-precision weapons by the defense industry?
Recall that in 2020, Russian President Vladimir Putin approved the national Strategy for the development of artificial intelligence for the next ten years. Speaking to scientists, developers and representatives of high-tech business, he said that it is extremely necessary for us to maintain leadership in this area. "Artificial intelligence is a resource of colossal power, whoever owns it will get far ahead," the head of state stressed.
To the credit of our advanced defense corporations (for example, such as KTRV, JSC NPO High-Precision Complexes, etc.), they have learned this well for a long time. Their products: the Iskander missile defense System, the Pantsir-S1 missile defense system, the Kornet ATGM, the Kalibr-NK, the Bal DBK, and the Bastion are now showing themselves in the best possible way in the combat zone. Largely because these products have integrated elements of artificial intelligence.
AFTER SWITCHING TO "DIGIT"
About two years ago, JSC NPO High-Precision Complexes prepared a forecast for the development of scientific and technical achievements until 2030 in this JSC and other holdings of the domestic defense industry.
The document was sent to the chairman of the NTS of Rostec State Corporation Yuri Koptev. It talked about innovation, including the creation of artificial intelligence. And, as we can see, this direction is especially in demand today.
"For specialists in the field of high–precision weapons, this is one of the most advanced technologies that will make a qualitative leap in the development of military hardware, since artificial intelligence radically changes everything," Nikolai Ivenev, adviser to the General Director of High-Precision Complexes JSC, told me at the time: "When processing signals, the use of AI or so-called deep ultra-precise neural networks makes it possible to significantly increase The efficiency of any product".
But before the full-scale introduction of sanctions by the West, some directors of enterprises were forced to buy foreign neural networks as a kind of "black box". So in the late 1980s we bought the first computers, often not understanding how they work.
However, sooner or later you have to pay for such "dependency". Therefore, serious scientists are trying to study foreign technologies, and then adapt them from us or create their own. Moreover, in foreign sources it is not possible to find, for example, how many layers in a particular neural network, why exactly such switching functions are in it, and not others, etc. If a neural network allows you to establish a multidimensional dependence of a function on a bunch of parameters, then what is the pattern of this process?
"We know that all neurons (an attempt to simulate the human brain) do not work by themselves, but are interconnected," Nikolai Ivenev confirms his words with an example. – There is a signal at the input, and a neuron at the output, which already has a certain function. It is possible to build such a neural network, because our microelectronics is quite strong, but the problem is different. How to calculate it, how many layers should there be in it, are there feedbacks in them?
In "High-precision complexes", several councils of chief designers devoted to this problem. The results of our developments can be applied in acoustics, in speech recognition, in obtaining information about a submarine in the optical and infrared ranges, in radar... So, when switching from an analog signal to a digital and phased array antenna arrays (FAR) and AFAR, the problem of angioech arises. In analog stations, all this was decided by our fathers and grandfathers, but not in digital radars. The neural network will allow us to solve these problems."
The topic of artificial intelligence for our defense industry is interesting when improving high–precision weapons - for example, when providing combat operations of robotic complexes in automatic mode.
Even under sanctions, this task can be solved. The availability of mathematical support is very important here. Therefore, specialists of the Keldysh Institute (RAS) were invited to one of the councils of the chief designers of JSC NPO High-Precision Complexes, since tensor (higher) algebra is used in these processes. We are talking about scattered matrices, their multiplication, addition, etc.
DISPARATE FORCES
And yet, frankly speaking, not all Russian scientists took this problem seriously – at least before the start of the special operation in Ukraine. Many have not decided on fundamental scientific directions.
So, out of 100 authors cited in the world in this field, there are three or four of ours. Almost all the others are English–speaking, although there are Chinese surnames among them. And what, for example, are the results of Rusnano's scientific activities over the past 15 years? Little is known about this. Although, of course, there are talented people there. One of the residents of this center, as I was told, is a young mathematician, Doctor of Sciences Ivan Oseledets. His works are cited abroad.
In the meantime, our state corporations are in many ways pioneers in the creation and implementation of artificial intelligence. For example, at the Mikhailovsky readings at the Central Research Institute of AG, young scientists made presentations with scientific reports. Representative of FSUE "GosNIIAS" (State Research Institute of Aviation Systems) Nikolai Moiseev raised the problem of "object recognition using ultra-precise neural networks."
The state needs to interact with such enthusiasts, help them master new technology, which is one of the most promising today.
MARCH OF ENTHUSIASTS
America took up neural networks back in 1943, when we were fighting on the Kursk Bulge. But after the Victory, we made a powerful breakthrough. Computers and supercomputers have already been laid in our "nuclear project".
There have been attempts to study what the brain is, how it functions. Let's recall the famous works of Natalia Bekhtereva. But in Khrushchev's time, failure came again – the fight against cybernetics as a pseudoscience began.
But the artificial brain is a neural network, although it cannot be said that the same processes are going on in the neural network. In the brain, physico-chemical and biological processes are associated with the transmission of information and its analysis. The neural network is an attempt to make some kind of analog of the brain using cybernetic systems.
There is a Foundation for Advanced Research and other scientific organizations in Russia. But there is nothing much to brag about here either – unless, of course, all developments immediately become top secret.
As experts have told me, it is extremely difficult even to create a hardware part of a neural network. To calculate a neural network, you need a supercomputer or a specialized computer – the so-called tensor processor that works with tensor algebra.
And yet we did something. Recently, for example, a similar processor, its topology, and architecture were practically created in Russia. But in metal, as it turned out, it is easier to produce it abroad – because we do not yet have the technology that would allow calculations at the level of, say, 28 nanometers.
If we manage, then in fact it will be a repetition of the Google neural network. But already Russian content, which is also not bad. When we have such a chip, a second task will arise: to make a hardware and software complex that will teach something. For example, to recognize faces, military equipment, etc.
I repeat, work in this direction is underway in the country – but, as often happens at first, mainly by enthusiasts. They don't even need much money. They just love what they do, they are fans in a good sense of the word. There are such, for example, in Zelenograd, near Moscow, at other Rostec enterprises.
Or take the KTRV corporation. As the general director Boris Obnosov told, the creation of hypersonic weapons in their scientific and technical problems was comparable to the creation of the first satellite. That is, it was extremely difficult. But this direction also has a multiplier effect, which is of great importance for the civil sector of the economy. These are materials science, engine building, aerodynamics, avionics, mathematical modeling – all at a new level.
The task of the management of any holding today is to support talented scientists and developers, unite them into a team, help build an experimental sample, test it, launch it into a series.
THE FATE OF THE PATENT
And there are many options for commercialization and implementation, up to the royalty agreement. Oddly enough, such a mechanism in the defense industry of Russia still practically did not exist. The state does not always immediately pay attention to this due to bureaucratic obstacles. And we need to think about this, too.
From now on, intellectual property must remain in the enterprise. It is wonderful. But we came to this relatively recently.
Previously, the Ministry of Defense, for example, set a lot of research work. Within its framework, one or two inventions were born, which were patented. But the development itself became the intellectual property of the Ministry of Defense of the Russian Federation (state). This has been the case since Soviet times.
And the people who developed the invention, implemented it, in fact received a bonus and... that's it. That is, the state simply appropriated intellectual property, and a person was given a piece of paper in the form of a diploma or certificate.
Now the situation has changed dramatically. For example, the legal department of JSC NPO High-Precision Complexes made a proposal to save and accumulate the results of intellectual activity at the enterprises themselves and in the holding. Moreover, many works are carried out in it at their own expense – which means that the very concept of "invention" acquires a different meaning. Patents can and should remain the property of the enterprise.
For example, the diploma was awarded to representatives of JSC "ZID" for the development of the topic "Method of creating topogeodetic networks for the preparation of combat operations of missile troops, artillery and air defense of Ground forces." Its relevance lies in the formation of special topogeodesic networks and the binding of geodetic reference points with the help of high–precision geodetic complexes. And the novelty is in improving the accuracy of space navigation tools used by consumers of GLONASS signals. Each of these inventions has its own patent, a specific authorship, which no one can encroach on.
And when an agreement is concluded with the Ministry of Defense of the Russian Federation, it stipulates in advance in what proportions the enterprise (holding) can claim its developments, be the rightful owner of the know-how produced. This, of course, stimulates the work of the inventor, scientist, innovator, and the collective as a whole. And if there is an opportunity to commercialize it, that's fine.
THE DIRECTION OF THE MAIN BLOW
By the way, our GLONASS system is no longer inferior to GPS, and in some ways it surpasses it. So disabling GPS will not affect the accuracy of targeting our "Calibers", planning ammunition and combat units, the actions of the troops as a whole. They both received and will continue to receive the most accurate information from our satellite constellation.
"In any holding there are many ideas, excellent developments that require systematization, combining them into promising innovative projects," Nikolai Ivenev reflects. – Therefore, it is necessary to boldly create structures such as the Scientific and Technical Committee (STC). You don't need a lot of money for this. We just need to gather a team of like-minded people and set them a task for, say, three years."
In any new product of the IWT, the main thing is an intelligent filling. Analog communication, for example, required a lot of time to develop and decrypt images from the same old UAVs like "Flight", "Swift", "Bee". And neural networks allow you to transmit ready-made information to the command post online, even in flight, up to target designation.
It is the branch research institute in the direction of high-precision weapons that should formulate and set the directions of such work. So, we need an institute of promising (innovative) developments of a holding company with a small staff and a research program. It's not about the name, you can call it an Innovative Technology Center, which should have three laboratories: for propulsion systems, artificial intelligence, communications. Their task is to generate ideas, engage in scientific, technological and technical search.
A TASK FOR GENIUSES
Surprisingly enough, but to formulate one or two serious, breakthrough scientific and technical directions is a task of incredible complexity. She is at the level of such talents and geniuses as designers Arkady Shipunov, Sergey Invincible.
What allowed them to make discoveries and breakthrough solutions? An unsurpassed talent. But in the USSR, the NTC of the glavkov (ministries) also worked for this, which solved the tasks of accumulating scientific and technical information. When conducting complex research, as a rule, it was formulated where to move on.
Now the absence of such structures makes itself felt. Although it is extremely important for any enterprise to find its own track, the branded product is a manufactured superweapon that all countries would like to have in their arsenal.
But to do this, you need to be able to formulate and set tasks correctly. The brand "Made in Russia" should be proud, associated with the strength, power and glory of Russian weapons.
Oleg Falichev
Oleg Valentinovich Falichev is a military observer.