The development of military artificial intelligence (AI) in Russia should reach a new level – a whole specialized department will appear in the Ministry of Defense responsible for the introduction of such technology. Why is artificial intelligence of fundamental importance for the military field and what kind of weapon will be equipped with it in the first place?
Until now, the development of “combat robots” was mainly constrained by the level of development of their control part – artificial intelligence (AI). The meaning of the appearance of a robot on the battlefield is quite understandable: if it is possible not to endanger the life of a serviceman, it is worth sending a robot to perform a complex task. Otherwise, it will turn out to be something like a Japanese aircraft-bomb “Oka” during the Second World War, where in each machine, instead of a control system, there was a kamikaze pilot with a one-way ticket.
The concept of robotization and “dehumanization” of military technology is by no means new – the first robotic systems, even with primitive actuators and controlled by a conventional radio signal or by using simple sensors, also date back to the Second World War. Here, too, the Axis countries distinguished themselves: one of the first real control systems was used by the German V-2 rocket. It allowed the “V-2” to hit targets at a considerable distance, which was an unpleasant surprise for the allies: after all, such a “deserted” weapon was a novelty then, and the “V-2” could inflict much more damage than the Japanese “Oki” …
The development of combat robots in our country also has a rich history. Tests of remotely controlled tanks in the USSR took place back in the 1930s, and the Soviet Union was in the lead in the production of combat drones and “smart” missiles until 1991.
Today we are no longer surprised by rockets that take off themselves, choose a course or even land – programmed actions for complex equipment have already become an everyday reality. Of course, it would be more correct to call such a technique, acting according to given programs, robotic systems, because the functions of making key decisions in this case are still more vested in the operator, that is, the person.
However, there is another fact: for example, even in the combination “operator – apparatus” existing in modern UAVs, many simple functions, such as adjusting the thrust or the position of the control planes, are performed by the robot, while the operator, for example, only sets the course and speed of movement cars. This approach allows the operator to unload as much as possible for making important management decisions, so as not to distract his attention with a mass of complex and non-obvious processes typical of any modern technology.
In addition, it should be understood that the operator of a UAV or a remotely controlled tank, sitting in a comfortable seat, does not “feel” the car in the same way as a pilot or driver. So the robot, willy-nilly, has to do it for a human.
What is the difference between a simple robotic technology and a completely autonomous one, which will be controlled by a real, “adult” AI? As a matter of fact, programmers, designers and engineers today need to go to the final goal quite a bit. Namely, to close the system of obtaining information from the battlefield to a complex decision-making apparatus, which, independently and autonomously from a person, will take the most correct response actions.
Such a device will not only be able to surpass the operator in the speed of decision-making, but will also be much more protected from enemy actions. After all, one of the main ways to combat unmanned vehicles is jamming or intercepting the control channel of the vehicle, after which the robot becomes either a “useless piece of metal” or, even worse, a “turncoat dummy”, which, for example, can be commanded to sit on the enemy’s territory …
Over the past decades, the United States has set the pace in developing AI systems for the military. There, the DARPA agency was responsible for promising systems, under the auspices of which, in recent years, even demonstration “competitions” of autonomous robots were held annually.
The tasks that were set before the participants of the competition were quite “peaceful”: the robot had to open the door, unscrew the crane, go through the obstacle course, lift the load, etc. The anthropomorphism of the robots gave additional entertainment: the “iron dummies” looked rather comical, resembling clumsy clowns trying to show a funny reprise in the circus arena. However, in such competitions, real systems were worked out: the work of machine vision, the creation of a dynamic 3D-picture of the environment, testing the system for making complex decisions, working out accurate control actions and feedback systems.
After 2015, further robot contests were held behind closed doors and under the auspices of the Pentagon, as their military focus became apparent. This suggests that the creation of fully autonomous combat systems in the United States has either reached the home stretch – or, at least, is not far off.
In Russia, such research is carried out by the Foundation for Advanced Study (FPI). One of his projects is a robotic platform included in the “outfit of the soldier of the future.” According to the FPI, the fighter should have a personal assistant – a robot squire. The complex can be considered as a soldier’s own “dog”, which will allow him to solve his tasks faster and more conveniently, transport his weapons and ammunition, provide communications, see farther and in different ranges, and also hit targets with his own weapons.
However, most likely, the first real autonomous systems will not be ground robots, but unmanned aerial vehicles (UAVs).
As stated in the Russian Ministry of Defense, the Defense Ministry plans to create a specialized department that will deal with the development of artificial intelligence. At the first stage, AI systems will be implemented in unmanned aerial vehicles.
This choice is due to the fact that, oddly enough, the control of the UAV and its environment are much more “analytical” than, for example, a battlefield for a wheeled or, moreover, a walking mechanism. In fact, the UAV moves in an empty and homogeneous space, in which the influence of the earth’s surface relief is insignificant, there are no “dead” zones in the surrounding space, and all objects, for example enemy aircraft, are easily distinguished from the surrounding background.
In addition, training a modern pilot is an extremely expensive process costing millions of dollars.
And the reaction speed of a person is finite: a person simply does not have time to react to a rapid change in the situation in the air, which the AI will successfully “work out”. Moreover, for the UAV operator, this will be further complicated by the signal delay, which still loses precious fractions of a second in the channels between the UAV and the control center. So the indicated speeds, at which the AI still manages to react to the change in the picture of the battlefield, have already become prohibitive for humans.
Of course, the UAV has its own difficulties: it flies itself at a speed of several hundred meters per second, and anti-missiles used against it can even reach speeds of 2-3 km / s. However, this determines exclusively the requirements for the speed of the controlling AI, while the behavior algorithms remain quite simple, and the enemy’s actions are predictable.
According to the concept of development and combat use of robotic systems, for the period up to 2025, the share of robots in the total structure of weapons and military equipment of the Russian army should reach 30%. A significant part of these systems will be able to become not only robotic, but also fully autonomous, operating under the control of artificial intelligence.