Military Review

Modernizing tanks with artificial intelligence: plans of the US army

24

In the US, thinking about the possibility of modernizing management tanks using artificial intelligence. Indeed, over the past 40 years, the methods of controlling armored vehicles have practically not changed, says analyst Nathan Stroot.


Tank management remained in the era of forty years ago


Despite the fact that the US military is trying to maximize the use of advanced technology in the military, a number of areas still use it to a minimum. First of all, these are armored units: Nathan Stroot writes that over the past forty years, tank control technology has changed little. While military engineers are thinking about creating robots, developing artificial intelligence, tanks are still operated manually by driver mechanics, and without a highly qualified "mechanic", a tank may be useless on the battlefield.

Sgt. Dustin Harris of the US Army says that even now, crews use a manual process to detect, identify, and engage targets. Tank commanders, driver mechanics, gunners - they all operate manually. Once the target is detected, ammunition is also manually selected to defeat it. All this, of course, takes extra time.

According to Richard Nabors, DEVCOM C5ISR's Acting Deputy Director for Systems and Simulation, the Army High Command is also aware that tank control technologies are outdated and have been functioning practically unchanged for forty years. Therefore, they attended to the integration of modern technologies into tank management and turned to the DEVCOM C5ISR center and the Command for the Development of Combat Capabilities for this purpose. And on October 28, the military invited the press to the training ground in Aberdeen to demonstrate the ATLAS improved guidance and fire control system.

As it turned out, the new product uses advanced sensors, machine learning algorithms and a new touchscreen display to automate the process of finding and hitting targets. As a result, crews can respond to emerging threats much more quickly than before. As project manager Dawn Deaver emphasized, the crews will be able to hit multiple targets during the time during which only one target is currently being hit.

Testing the new system


Nathan Strote compared the ATLAS prototype to something from the Star Wars movie. The system was mounted on a sleek black General Dynamics Griffin I chassis with a 50mm automatic loading turret from the Army's Advance Lethality and Accuracy System for Medium Calibur (ALAS-MC). A small circular Aided Target Recognition (AiTR) sensor was mounted on the turret. It is a medium wave infrared image sensor that scans the battlefield.

Behind the tank was the M113, connected by a series of long black cables. During the demonstration, the crew was located inside the M113. Of course, we cannot yet say that a new tank has been developed. The demonstrated mechanism is a prototype of a future tank or combat vehicle, into which artificial intelligence will be integrated. ATLAS is independent of the chassis or tower on which it is installed.

When the target detection process is completed and the target is already in the scope and selected weapon to defeat it, the operator has a choice: approve the AI ​​recommendations and pull the trigger, adjust the settings before pulling, or abandon the fight. The whole process from target detection to the start of fire takes seconds. After destroying a target, the operator can simply touch the touchscreen to select the next target detected by the ATLAS system.

Beginning in August 2020, the US Army is recruiting small groups of tankers to test the new system. Collecting feedback from the soldiers and sergeants themselves, military engineers expect to be able to quickly improve the system and prepare it for real use in combat conditions.
Author:
Photos used:
Twitter / US Army
24 comments
Information
Dear reader, to leave comments on the publication, you must to register.

I have an account? Sign in

  1. Lexus
    Lexus 30 October 2020 16: 16
    15
    Neural networks are no longer the future, but harsh everyday life.
    1. pytar
      pytar 30 October 2020 16: 47
      +2
      Neural networks are no longer the future, but harsh everyday life.

      Yes, Alexey! The future has come and is unknown, is it good or bad ?! request
      1. Lexus
        Lexus 30 October 2020 16: 52
        12
        "An autopsy will show." (FROM)
        Or maybe already excavations.
        1. pytar
          pytar 30 October 2020 18: 04
          0
          More likely - "evolution" goes in circles! laughing

          Robots will become smarter than humans, and in a moment they will say - "Why do we need this stupid soft body?" bully
          1. vVvAD
            vVvAD 30 October 2020 19: 51
            0
            Have you seen enough of Terminator and The Matrix?
            First, create a machine that learns faster than a human, and then we'll see.
            There are great doubts that this is possible in principle.
      2. vVvAD
        vVvAD 30 October 2020 19: 49
        -2
        It is known that such systems are present in the latest domestic air defense systems. And, as you know, they are the best in the world. bully And, presumably, they are implemented in the "Armata" family.
  2. rocket757
    rocket757 30 October 2020 16: 33
    +4
    Neural networks, AI ... so far, only an imitation of the natural. Over time, as this technology develops, everything will become more complex and interesting.
    Oh yes, the question is, but .... they will replace the big loader with what / who?
    1. Nasdaq
      Nasdaq 30 October 2020 17: 36
      +3
      Oh yes, the question is, but .... they will replace the big loader with what / who?

      A smart charger (an IT specialist who was not hired by an IT company) with virtual reality glasses laughing
      1. rocket757
        rocket757 30 October 2020 18: 55
        +1
        Ha, then the ammunition will also have to be made virtual and it will hit the virtual enemy .... although we have a couple of not small guys in the department! Not two-meter blacks, but the same size is nothing.
  3. Operator
    Operator 30 October 2020 16: 58
    -4
    Computer recognition of an openly located target is already seeds.

    The Australians have made their system even cooler - the computer tracks the movement of the tanker's eye across the field of augmented reality glasses and determines the places where the eye lingers a little more, and then displays a magnified image of these places on the screen for the tanker to examine in detail to identify the camouflaged target.
    The Australian human-machine system makes optimal use of the tanker's experience and the capabilities of the computer.
    1. APASUS
      APASUS 30 October 2020 19: 14
      +2
      Quote: Operator
      Computer recognition of an openly located target is already seeds.

      The Australians have made their system even cooler - the computer tracks the movement of the tanker's eye across the field of augmented reality glasses and determines the places where the eye lingers a little more, and then displays a magnified image of these places on the screen for the tanker to examine in detail to identify the camouflaged target.
      The Australian human-machine system makes optimal use of the tanker's experience and the capabilities of the computer.

      Of course of course .............
      Checking the adequacy of training
      Even in the case of successful, at first glance, learning, the network does not always learn exactly what the creator wanted from it. There is a known case when the network was trained to recognize images of tanks from photographs, but later it turned out that all the tanks were photographed against the same background. As a result, the network "learned" to recognize this type of terrain, instead of "learn" to recognize tanks. Thus, the network “understands” not what is required of it, but what is easiest to generalize.
      Testing the quality of training a neural network should be carried out using examples that were not involved in its training. Moreover, the higher the quality of training, the greater the number of test cases. If neural network errors have a probability close to one billionth, then a billion test cases are needed to confirm this probability. It turns out that testing well-trained neural networks becomes a very difficult task.
      1. Operator
        Operator 30 October 2020 20: 16
        -3
        I'm not talking about neutron networks, I'm talking about the software of your smartphone, which distinguishes the faces of people in the frame, and in some models, the user's face as a password.
        1. APASUS
          APASUS 30 October 2020 20: 33
          +1
          Quote: Operator
          I'm not talking about neutron networks,

          AI is not possible without the development of neutron networks, as I understand this material
          1. Operator
            Operator 30 October 2020 20: 45
            -3
            AI is not needed for target recognition.
  4. Military77
    Military77 30 October 2020 17: 31
    +2
    Isn't that the poor Negro loader at the AZ going to be replaced? Or will AI control and charge the Negro?
  5. Azimuth
    Azimuth 30 October 2020 18: 28
    +1
    Quote: Operator
    Computer recognition of an openly located target is already seeds.

    The Australians have made their system even cooler - the computer tracks the movement of the tanker's eye across the field of augmented reality glasses and determines the places where the eye lingers a little more, and then displays a magnified image of these places on the screen for the tanker to examine in detail to identify the camouflaged target.
    The Australian human-machine system makes optimal use of the tanker's experience and the capabilities of the computer.

    Augmented reality glasses were tested several years ago by European companies for mechanics and vehicle commanders, according to the results, optoelectronic systems with a thermal imaging sight and dubbing with conventional optics were recognized as the most reliable.

    From the fresh look on the main topic: https://youtu.be/nwvxVFb6Rzc
    1. Ua3qhp
      Ua3qhp 30 October 2020 21: 47
      0
      Quote: Azimuth
      Augmented reality glasses were tested several years ago by European companies for mechanics and vehicle commanders, according to the results, optoelectronic systems with a thermal imaging sight and dubbing with conventional optics were recognized as the most reliable.

      Yes, while the matrices for glasses are imperfect, too large pixels and some other features when the screen is close to the eyes. But they are working on this, both civilian and military. Over time, this will be the main screen, both on armored vehicles and on individual infantrymen.
  6. tank64rus
    tank64rus 30 October 2020 18: 32
    0
    So far, this is just the beginning, and perhaps with a commercial implication, but it is unwise to discount all this.
  7. Azimuth
    Azimuth 30 October 2020 19: 41
    -1
    Quote: tank64rus
    So far, this is just the beginning, and perhaps with a commercial implication, but it is unwise to discount all this.

    Only in the comfortable struggle of the white Sahibs against the natives, so that a tank with a refrigerator full of cola and a toilet, in a major regional conflict and even more so a global war where charges "with power in TNT equivalent" will be used, is the price of all this electronics a penny.
  8. IS-1
    IS-1 30 October 2020 19: 51
    -1
    Modernizing tanks with artificial intelligence: plans of the US army

    Anglo-Saxons just not to fight directly, but only to play off ..))) Are they hoping for artificial intelligence? lol
    This will destroy them sooner or later .. What are they trying to achieve with this persecution and threat to Russia?
    In the United States themselves, there are pogroms and blood, and in Europe ... and everything is a rod for Russia))))
    The last chance to survive and preserve capital and influence? Oh well
  9. Gunther
    Gunther 30 October 2020 22: 13
    +3
    Yankees as always at the cutting edge of progress - now will be charged muscular black man with n (r) eyroshuntom in the ass, under the control of artificial intelligence. belay
    lgbt community of the usa think will approve ...
    we are empty better the old fashioned way - the machine is charging soldier
  10. psiho117
    psiho117 31 October 2020 02: 35
    0
    The only strange thing is that no one has said "wow, EMP will shut down all this high-tech crap of yours."
    And also electronic warfare suppresses 2 Cooks per minute.
    wassat drinks
    1. Mavrikiy
      Mavrikiy 31 October 2020 08: 25
      0
      Modernizing tanks with artificial intelligence (AI): plans of the US army
      An attempt to make candy out of the Abrams shit, albeit at a cost. Putting sensors instead of levers is ingenious, but not professional. Some UAV models showed the capabilities of AI in Karabakh, and in the tank there is an order of magnitude more work for it. And yet not everything is so simple, both electronic warfare and reliability. Weapons must be guaranteed to fire when needed. The UAV refused, they launched another, this one for maintenance, and the tank ...
  11. iouris
    iouris 31 October 2020 09: 36
    0
    Finally, we have created a "homo unintelligent". AI to help him!