Since we've been in orwellian times words don't mean a damn, I know. But FWIW you don't mean "Ukrainians" I'm quite sure, you mean 'Kiev Ukrainians' in your own mind and even there you are wrong for you are really referring to the outcome from the collaboration of the best minds of 50 nations. Including some of the most advantaged nations in the world. It'd be strange the, if they didn't come up with something, really, wouldn't it?
Yes, but be aware the Russia is also innovating here and while I don’t think they can manage the speed and breadth of Ukrainian innovation they are still capable of finding new ways to hamper as well as new ways to damage. But yes, it is impressive of Ukraine. No we just need to learn.
Training design/layout of logic circuits on FPGAs instead of using neural networks has been shown to yield considerable performance-to-computing-power improvements with ATR, but it's even harder than with neural networks. Note that NVidia themselves have used this approach to come up with more efficient GPU designs. Perhaps this needs a two-tier approach - evolve the FPGA circuit, and optimize/evolve the evolver at the same time. If Ukraine is able to pull this off, it could see ATR becoming mainstream in small devices.
I am anxiously waiting for someone to conceptualise these very long-range targeting complexes as Human-In-The-Loop-Extended-Range systems and then struggles with the acronym. Apologies, I have nothing serious to add to this excellent article which I thoroughly enjoyed and am very grateful for.
Attacking the rail system would seem a straightforward task for ATR. Routes are finite and easy to locate, and recognizing locomotives would be a narrow task
Also fits with what I have learned, still I think we are far away from a breakdown in logistics. Poorer, indeed, but not breakdown. Still let’s hope someone does a knowledge based analysis of your proposal. My reasoning isn’t necessarily strongly data based.
Nice article, clearly explained and still good for the HITL target discrimination control being the primary control before terminal hit for the Ukrainian drones. Thanks for sharing Tom.
The development of hardware and models is continuously evolving. Additionally, new offensive/defensive strategies are being developed to achieve the best possible success against drones.
It is not far off when I say that in the future these tactics with swarms will be carried out. Swarm communication (drone to drone) and also swarm targeting, which is based on already known infantry procedures. When uncertainties arise, the number of drones involved will increase so that a better decision can be reached than just one drone alone. I see potential for further innovation in the near future, so I also say: "Don't say 'Never'
Fascinating! I’m Harrison, an ex fine dining industry line cook. My stack "The Secret Ingredient" adapts hit restaurant recipes (mostly NYC and L.A.) for easy home cooking.
It's amazing how the Ukrainians managed to become arguably the most advanced military regarding drones.
Since we've been in orwellian times words don't mean a damn, I know. But FWIW you don't mean "Ukrainians" I'm quite sure, you mean 'Kiev Ukrainians' in your own mind and even there you are wrong for you are really referring to the outcome from the collaboration of the best minds of 50 nations. Including some of the most advantaged nations in the world. It'd be strange the, if they didn't come up with something, really, wouldn't it?
Sure, tovarisch, only glorious muscovy can develop their weapon systems relying on own SoC/SoM production.
Unlike us "kieff" Ukrainians who are basically mindlessly drooling over tech coming from 'most advantaged nations.'
Is that so? I had not known. Thank you for the headsup. :)
The hell are you yapping about?
Yes, but be aware the Russia is also innovating here and while I don’t think they can manage the speed and breadth of Ukrainian innovation they are still capable of finding new ways to hamper as well as new ways to damage. But yes, it is impressive of Ukraine. No we just need to learn.
Yes for sure!
Training design/layout of logic circuits on FPGAs instead of using neural networks has been shown to yield considerable performance-to-computing-power improvements with ATR, but it's even harder than with neural networks. Note that NVidia themselves have used this approach to come up with more efficient GPU designs. Perhaps this needs a two-tier approach - evolve the FPGA circuit, and optimize/evolve the evolver at the same time. If Ukraine is able to pull this off, it could see ATR becoming mainstream in small devices.
I am anxiously waiting for someone to conceptualise these very long-range targeting complexes as Human-In-The-Loop-Extended-Range systems and then struggles with the acronym. Apologies, I have nothing serious to add to this excellent article which I thoroughly enjoyed and am very grateful for.
Attacking the rail system would seem a straightforward task for ATR. Routes are finite and easy to locate, and recognizing locomotives would be a narrow task
Absolutely. Just find the track and follow it until you hit something.
But is it really worth it? Relatively easy to repair I think. And there are lots of it.
From what I have read elsewhere Russia is dependent on rail and has an increasing shortage of locomotives.
Also fits with what I have learned, still I think we are far away from a breakdown in logistics. Poorer, indeed, but not breakdown. Still let’s hope someone does a knowledge based analysis of your proposal. My reasoning isn’t necessarily strongly data based.
Nice article, clearly explained and still good for the HITL target discrimination control being the primary control before terminal hit for the Ukrainian drones. Thanks for sharing Tom.
The development of hardware and models is continuously evolving. Additionally, new offensive/defensive strategies are being developed to achieve the best possible success against drones.
It is not far off when I say that in the future these tactics with swarms will be carried out. Swarm communication (drone to drone) and also swarm targeting, which is based on already known infantry procedures. When uncertainties arise, the number of drones involved will increase so that a better decision can be reached than just one drone alone. I see potential for further innovation in the near future, so I also say: "Don't say 'Never'
Thanks for the article. Very interesting. Any thoughts about the LSE report on drones? https://www.lse.ac.uk/ideas/Assets/Documents/Research-Reports/2025-04-05-DRONES-MatlackSchwartzGill-FINAL-WEB.pdf
Fascinating! I’m Harrison, an ex fine dining industry line cook. My stack "The Secret Ingredient" adapts hit restaurant recipes (mostly NYC and L.A.) for easy home cooking.
check us out:
https://thesecretingredient.substack.com
Great article! Complex technical information explained step-by-step.