Hacker News Reader

Comment by AHOHA | Hacker News Reader

While I’m working mainly on drones/robotics, counter drones is also becoming a big talk with a lot of interest from investors, however, countring drones isn’t that easy, even with latest sensors (sound/visual/Lidar/radar sensors), there’s a lot of false positives when deployed in real scenarios, one trial we tried near an airport, the system we tested caught a lot of birds as drones, you tune it not to detect it, then drones goes undetected. Some systems tries to us AI to study the fly path and find some anomalies and trigger based on that, but so far I didn’t see personally an acurate system that can reliably detect and neutralize. Swarm on the other hand, might be easier given how easy to detect them.

View on HN

There's a lot of electronics on a little drone that all produces a lot of noise. Depending on the control schema it may be dumping a video feed or telemetry to the base station (typically over 2.4Ghz?). There's also an incidental EM footprint put out by the electronics on board. Brushless motors are extremely electronically noisy and shielding is expensive (by weight, aluminum foil works, but to be effective it needs to be wrapped around everything). I'd go hunting for ways to characterize that EM signature at a distance. Do you need to point a big collection dish at the target? Yes, very likely, but it should get you beyond the standard set of false-positive scenarios caused by Not Enough Information (TM).

Can we listen for motor noise broadcast by the brushless motors?

> Brushless motors are extremely electronically noisy and shielding is expensive

If we're comparing RF noise quantity, brushed motors generate a heckuva lot more RF because they're spark generators. Of course modern drones don't use brushed motors much.

Brushless motors can in principle have their RF noise reduced by careful modification of their controller waveforms.

But your point is well taken. Brushless motor RF is probably easy to identify by the presence of a few peaks on the spectrogram representing multiples of the rotation frequency. Brushed motors would look like that too except buried in a lot more white noise.

C2 (command and control) and Telem links can be communicated over different means, either cellular, satcom, or even LoRa, how you will differentiate the drone modem from thousands of UEs in the vicinity? what if the drone flying at a low alt, near power lines and other EM noise? one time I made a drone that detect interference sources for a gov agency to detect any violation of the spectrum, a giant LED can cause so much noise more than any brush-less motor, let alone a small drone ones. Maybe if you have a giant drone (plane size) and flying at a high alt in an empty area, maybe, anything else I would say it's impossible or very hard to reliably use EM as way for detection.

> there’s a lot of false positives when deployed in real scenarios, one trial we tried near an airport, the system we tested caught a lot of birds as drones

Those weren’t false positives.

https://birdsarentreal.com/

Airport is probably a very different environment, though? I imagine that if you have a military installation, you don't mind blasting some birds with RF every now and then (and the birds probably don't mind either).

As a passenger on commercial aircraft, I don't mind reducing the bird population near an airport.

A bird is small and largely non-conducting. I'd guess it could plausibly survive a burst of microwave that could disable a drone, at any practical distance from the weapon.

Microwave tends to "cook" you.

Microwave covers a huge range 300 MHz to 300 GHz and they want something that will work in the rain or fog which means minimizing absorption by water as much as possible.

I think that if a drone is threatening civilian airline traffic you don't mind blasting some birds, either.

Unfortunately I do mind. Here’s where I live, there’s a bunch of endangered birds that live near the airport and don’t live literally anywhere else. Can’t really just go shooting them with death rays.

If any birds are in that situation near a military base or war zone I hope we got some if them in a zoo or dna as a backup for if we invent really good cloning since otherwise those birds are extinct

There's no war zone near me, so I guess the military doesn't need to shoot birds.

unfortunately, Korea have

The priorities at that point would probably be stopping any terrorist threat before protecting endangered birds.

There's no terrorist threat but there are birds, so we shouldn't just shoot birds for no reason.

Same here. I have zero terrorists in my backyard. I do have kids, rabbits, lots of birds, kites, etc. What's up with these idiots thinking about putting this microwave bird killer in my back yard where there are no threats against which to protect. Idiots.

ok_dad's Airport Cafe

Special tonite only

(all with salad & fries):

- roasted pigeon $3.99

- roasted quail $4.99

We cater to large groups!

Silly question but is it possible that the army would not care about birds triggering false positives if the weapon can do the work fast enough?

Yes. You give away information about your position and capabilities if your defensive system triggers inadvertently.

How about trying your tech in Ukraine.

> Russian electronic warfare (EW) remains potent, with an approximate distribution of at least one major system covering each 10 km of front. These systems are heavily weighted towards the defeat of UAVs and tend not to try and deconflict their effects. Ukrainian UAV losses remain at approximately 10,000 per month.

Does not matter. Total bird death. We need a system like that anyway for mass bird killing to actually contain avian influenza, otherwise it's just a game of waiting until it gets into humans.

The best way to detect UAVs remains the EM they emit themselves, be it telemetry or electromotive hum.

The footprint for small drones is almost negligible, and flying at lower altitudes make it even harder to detect.

Shoot a bird laser [1] at it and if it doesn’t respond, it’s a drone.

[1] https://www.birdcontrolgroup.com/

For the motor hum, yes. Telemetry/Control however is very easily detectable by design alone. Especially for consumer drones and their known spectrum/patterns.

>Telemetry/Control however is very easily detectable by design alone.

Is it? as I said in another comment, I would love to know how you will detect a drone C2/Telem link flying over cellular, or LoRa?

Consumer drones have RC links, flying either at 2.4GHz or ~900Mhz, that can be easily detectable, but you can easily equip a consumer one with other communications capabilities

How hard is it for a drone to detect your counter drone system and target it?

I guess if they managed to miniaturize it so that it was handheld somehow? Seems a bit performative but I guess lots of stuff did 20 years ago.

but drones are cheap, so make them sacrificial to find the counter-drone battery and then send in firing mission on the battery

That's what I mean by performative, yes.

If it were somehow miniaturized and handheld then it's easy to move.

I doubt that's physically possible, but I've been surprised before and am old enough to have been surprised before about how clever people are.

It looks like they are primarily thinking of this as protecting a non-mobile installation like a base vs troops on the move. So miniaturizing for hand-held might come, but the power supply for that would also have to shrink making it less useful. I see it mounting to a truck as part of the convoy moving troops during an advance to make it mobile

I would be extremely impressed with the engineering that would be required to somehow overcome the issue of focusing that wavelength at range. They don't mention the range in the article, so I'm suspicious.

It's extremely hard to imagine it having a range bigger than that which is effectively covered by other methods already. The laser makes way more sense to me, at least then the energy isn't spreading out super fast. But you never know, I absolutely did not believe optics would get sophisticated enough to manage realtime actual 3D displays (not VR, like https://voxon.co/technology/).

If they can do that with optics, I know I can't imagine what people with giant budgets can do with antennae.

Maybe a muon or electron or some type of particle beam weapon may be a better idea. Although maybe the wave vs particle spookiness may rear its head and a Faraday cage may stop one of those things too.

Do you mind sharing what make (vendor) of radar being used? The “AI” in these types of systems is usually more likely classical ML than deep learning, in my experience. The problem is solvable, but it needs more training data and investment to solve to some acceptable performance level. There are a lot of companies trying to cash in on counter drone, and hype-speak like “AI” abounds.

A spy!8-(

Reply on news.ycombinator.com