Comment by tamimio
9 hours ago
And that’s ok if it’s failing to do the job as intended, learning is acquired, and it looks fun to build, I am in the field and I find it great homemade concept.
Realistically, I doubt there’s ANY system out there will be able to counter small weaponized drones that are flown manually let alone with AI, you might have some workarounds, but never a real counter.
> Realistically, I doubt there’s ANY system out there will be able to counter small weaponized drones that are flown manually let alone with AI, you might have some workarounds, but never a real counter.
Weaponized drones (say D_A) can be countered by other weaponized drones (say D_B), equally cheap or cheaper than D_A because the D_A is usually targeting something larger (so more payload) and typically has a longer range. D_B only needs to wreck D_A at a shorter defensive range. That's what Ukraine is doing.
You can also use drone swarms with coordinated action so that each drone in the swarm is only targeting one other drone, and automatic re-targeting if one node misses. [1]
[1] https://en.wikipedia.org/wiki/Swarm_robotics
> equally cheap or cheaper
I doubt it, as D_A's target is stationary (and could be reduced to GPS coords) while D_B's target is moving.
> I doubt it, as D_A's target is stationary (and could be reduced to GPS coords) while D_B's target is moving.
It's a good point, though I should point out that GPS denial is assumed in those sort of contexts as a first countermeasure so D_A likely has alternative targeting, and that smaller drones can move faster with less energy storage, which itself requires less weight, compounding the benefits of being smaller.
And also the attacker can send 100 drones without any real targeting at all and 10 proper expensive drones and you need to send up 110 defenders which need to be able to track flying drones. Being the attacker will always be easier.
2 replies →
However, D_A is moving, while D_B can be stationary.
1 reply →
Couldn’t post earlier seems HN is rate limited :/
hardest issue as I mentioned in another comment is detection. Now on using other drones to counter a drone, there are other issues, as I built and tested some before, assuming you got the detection part done. The first one is guidance and correction mid-air, flying manually won’t really be practical due to the need for an extraordinary flying skills, which can’t be relied on in the field, the second part is the speed, you need to ALWAYS make sure the interceptor is faster to catch it up, third is the weight, I disagree about the payload part you mentioned, I have seen videos of light weight drones failing to wreck bigger ones, if you are relying on collision alone. Additionally, the telemetry/video/C&C for the interceptor, if jamming is already in place, your counter won’t work either.
The swarm will require a low latency comms link, centralized or decentralized, if the area is jammed, it won’t work. i have built a self-healing decentralized system using cellular in each drone, but that’s useless if the network is down to start with.
So they might work in a very specific use case, but not an ultimate solution to counter them.
While it [1] doesn't talk about swarms, it has some details - $1k - $2.5k price, 170mph speed, backpackable, thermal imaging, radar, ai, manual control (fiber-optic I think, based on other sources and battlefield pictures).
This [2] talks about swarmer software used by Ukraine.
$1k-$2.5k gives a lot of room for tech to avoid jamming - ir or visible light, ultrasound, for in-swarm comms.
And I wonder if the battery itself could be weaponized. We have seen that a very thin layer of the right material can turn phones/pagers very destructive.
[1] https://www.twz.com/news-features/ukrainian-companies-prohib...
[2] https://united24media.com/war-in-ukraine/ukrainian-drone-swa...
I don’t know if it will work, but here’s a startup that seems to be building an AI-controlled shotgun:
https://www.ycombinator.com/companies/9-mothers-corporation
Given the war in Ukraine, wanting to build such things is certainly understandable. But still, this is the stuff of nightmares.
David Suaez in Kill Decision had swarms of small single shot drones with the targeting intelligence of ‘00 camera. Identify a face, fly towards it, fire when close. It was an implementation of quantity has a quality all its own.
"Realistically, I doubt there’s ANY system out there will be able to counter small weaponized drones that are flown manually "
Why would lasers not work?
Those cheap drones are made from plastic, if you have a laser powerful enough and a target guidance system (like a camera and a PI) - then you would just need enough of them.
At long distances the small cross section of the drone requires tight focusing (expensive optics) or a high power, preferably pulsed laser (expensive laser) or both.
Not impossible but many times more expensive than the drone
Expensive is fine since it is reusable.
3 replies →
At very short distances and with a lot of power, perhaps. Despite what we see in movies laser beams diverge. And then with distance it’s harder to track moving objects precisely to hit the same spot long enough to melt it.
At that point might as well spend the money to use a kinetic weapon with basic tracking and ballistic calculations.
Kinetic weapons pose greater risk to bystanders.
The practicalities of using lasers are covered in some depth on the Naval Gazing blog. First part here:
https://www.navalgazing.net/Lasers-at-Sea-Part-1
Powerful enough laser and accurate enough targeting system is easy to say, but not easypeasy to do. Dumping thousands of Joules on a tiny moving target is much easier to do with explosives.
Lasers imo don't really have IRL advantages over machine guns and rockets, and their line of sight nature is a huge limitation.
Laser:
- are cheap to shoot - do not fall on someones head if they miss (unlike firing bullets and rockets at a drone that will come down again) - do hit the target immediately if aimed right
Problems with lasers are, cooling, power consumption limiting mobile use - and indeed targeting and fog and clouds.
1 reply →
Lasers won’t effectively work, it’s a two part equation, detection and targeting. To neutralize a target using a ground-based laser, you need an enormous power, and still it won’t penetrate a high distance/altitude in the sky, environment factors also to be considered. The detection part is even harder, these small 8in drones are almost impossible to detect unless you can hear it, aka it’s over, because they can fly at 250km/h, too small to be visually detected, acoustic sensors will fail to detect them, and radar will miss it as a false negative since it’s the size of a bird. I have seen some systems trying to combine all that to detect them plus AI for flying pattern detection, but they are far from being reliable in practical applications.
8 inches drone cannot carry much of explosive at all. In order to dump 10 kg load of explosives, you need an “agricultural drone” one that can carry 45kg, since the additional mechanisms and their batteries (and the drone’s backup batteries) are heavy.
Those are bigger and noisier.
DJI ARGAS Series are good starting point.
https://ag.dji.com/mobile
2 replies →
Unless you mean it just can't detect objects that small, my guess is we'll see things calibrate toward a lot more birds being cooked in active war zones vs drones with explosives being let through.
Can radar distinguish from the bird since it’s moving 250km/h?
The small weaponized drones do not fly 250 km/h.
2 replies →
From what I can tell, Ukrainians are having some success with converting guns into automatic turrets that can track and shoot down drones via sensors, and the rifle-equivalent of birdshot.
"let alone with AI" what's falling into the AI category here perhaps is the key question, since microseconds counts, and LLM are very slow!
Even the fastest "real-time" LLM frameworks currently report sub-second latencies around 120ms. This is fine for high-level mission planning (e.g., "fly to the red house") but too slow to prevent a drone from hitting a tree at 50mph (80 KM/h)[1]
Whilst the Shahed-136 kamikaze drone typically flies at a maximum speed of around 185 km/h (roughly 115 mph or 100 knots).
[1] https://arxiv.org/html/2602.19534v1 [2] https://en.wikipedia.org/wiki/HESA_Shahed_136
> "let alone with AI" what's falling into the AI category here perhaps is the key question, since microseconds counts, and LLM are very slow!
LLMs (Large Language Models) are far from the only type of AI around. It's a pretty broad field, and there are real-time AI systems, for example, self-driving cars, which have the response times you're thinking of. [1]
[1] https://en.wikipedia.org/wiki/Artificial_Intelligence:_A_Mod...
hence: "what's falling into the AI category here perhaps is the key question"
YOLO is a good example for something that can work.
https://docs.ultralytics.com/models/yolov8/
> Realistically, I doubt there’s ANY system out there will be able to counter small weaponized drones that are flown manually let alone with AI
What kind of systems are you thinking about? Jet airplanes for sure are completely safe from small drones.
"Jet airplanes for sure are completely safe from small drones."
That feels like a bold and unsupported assertion. Ask a pilot how they'd feel about takeoffs or landings through airspace filled with adversarial drones.
A flock of unlucky geese can knock out a jet turbine, how is this a "for sure" conclusion?
> Jet airplanes for sure are completely safe from small drones.
Until they land then, due to their cost, they become a very juicy target to aim for.
maybe in the air, but I seem to recall the Ukrainians being successful at attacking Russian planes on the ground.