Drones and how to kill them?

[VTOLicious]

There is no small/light weight/inexpensive INS that would allow precise navigation in a GNSS denied environment.

 

[Orionblamblam]

There is no small/light weight/inexpensive INS that would allow precise navigation in a GNSS denied environment.

Flies, bees, hummingbirds, pigeons manage it. Nothing special about them that AI won't eventually figure out.

 

[VTOLicious]

There is no small/light weight/inexpensive INS that would allow precise navigation in a GNSS denied environment.

Flies, bees, hummingbirds, pigeons manage it. Nothing special about them that AI won't eventually figure out.

I agree, maybe someday in the future. However, at present such technology does not exist.

 

[Orionblamblam]

There is no small/light weight/inexpensive INS that would allow precise navigation in a GNSS denied environment.

Flies, bees, hummingbirds, pigeons manage it. Nothing special about them that AI won't eventually figure out.

I agree, maybe someday in the future. However, at present such technology does not exist.

The future is coming fast. Just ask all the artists who are starting to realize that the days of actually getting paid to create art are coming to an end, because the AI can create equivalent art that's adequately good, vastly cheaper and vastly faster. AI can fly planes, drive cars, walk robots and hold conversations. Five, ten years down the line, AI capable of flying bird-like drones through bird-like behaviors seem not at all unlikely. Given that there is much more of the future than there was of the past, unless technology grinds to a halt there's no reason to assume that anything not forbidden by physics will fail to come to pass.

This stuff, like designer diseases and designer babies, is coming. It's not just insane but criminally negligent to not start planning for it.

 

[Kat Tsun]

If Javelin swarms aren't a thing it's hard to believe that "drone swarms" will become a thing anymore than waves of cruise missiles are a thing.

How about "cell phone swarms?"

The complexities of the anti-tank missile aren't huge, you're right. The expensive electronics can be replaced, and they have been in the latest Javelins, by modern electronics. The expense comes from the warhead, the sensors, and the motor. An electric drone can eliminate the cost of the motor but at the expense of mass, although this gives it some loiter time. They're still rather fundamentally different weapons and they can't replace each other, at least not yet. The electric propeller missile is still too slow and too large to be easily man portable with a similar punch and sensor capability as the solid fuel rocket.

Anyway yes as I said the biggest threat a "drone swarm" produces is a large quantity of flying hand grenades that go about a mile or so. Maybe two or three. Bear in mind that the T-62 was considered by the US Army to require about 10 DPICM bomblets/40mm grenades to ensure defeat, because HEAT charges are rather pathetic in behind armor effect, especially at small calibers. So it's effectively a mobile form of DARPA's self-healing minefield. A mobility obstacle for sure, something to be avoided or bulled through as necessary, but not a particularly lethal or dangerous threat compared to anti-tank missiles. It will require some measure of platoon or individual level air defense but it's not a particularly dangerous problem.

The more dangerous threat has existed since the 1980's: near real time satellite surveillance of ground forces. US SECDEFs had the ability to track mobile armored divisions using DSP satellites and the US Navy was interested in leasing DSP for warning to carriers against supersonic bomber attack. It's very hard to conduct offensive operations if your every move is anticipated and communicated as it happens.

Reconnaissance, at least in the traditional sense of finding the enemy in the field, is slowly becoming a thing of the past.

Drone swarms are rather agnostic to this because they don't materially offer any wide area surveillance or real time intelligence gathering that isn't better done by a high altitude reconnaissance aircraft like a NRO satellite or widely spaced magnetic/acoustic sensors like FCS was supposed to get. Individual drones, though, like the Black Hornet Nano, are probably quite lethal in their ability to provide real time intelligence with near zero signatures.

The problem with drone swarms is that people assume they are little quadcopters with Javelin warheads. They're not going to be that. They might be big quadcopters with hand grenades. Or actual airplanes with RPG rounds. Or a propeller powered cruise missile. The payload fraction is still in favor of the solid fuel rocket, which is the preeminent form of delivering chemical energy to a target and isn't going anywhere.

OTOH "cell phone swarms" can be effective as a means of visualizing flash mobs of angry people armed with robust mesh networks, homemade explosives, and craft produced firearms. Or, if we're sticking with cellphones, they could be armed with software defined radios, script kiddie jamming systems, and encrypted communications. That would be an annoyance to any combined arms force. That's a real futuristic threat to military forces, as the conventional "boots on the ground" are still slowly congealing into strategic police forces rather than actual combat troops. That's a threat for the future to realize though.

Cell phones are mostly a liability right now though. They present near idiosyncratic, personal identifiers of ground troops that can be easily coordinated by even 1980's-vintage field forces (the Russian Army) into lethal effects with the proper training and understanding of the technology, much less surpassing it. Perhaps they would be useful for guerrillas or insurgents but I'm not so sure. The US was rather effective at deploying tactically and operationally relevant anti-telephonics jammers in Iraq.

Practically speaking I think the drone swarm is much less interesting than the possibility of increasingly small and increasingly light high powered sensors. Drone swarms are big, like a flock of starlings, and can be easily tracked and avoided. Thus, their only real use is channeling a mobile force that can't protect itself from them, rather than defeating it entirely. Hence, mobile minefield.

Cell phones so far haven't shown anything much of merit on the battlefield. They may yet, since mesh networks are robust, and cellphones are really hardy. Hardier than military radios that's for sure, at least against nuclear SREMP and possibly TREE effects, because they're more modern electronics. That's for the future to reveal.

A tiny bird-sized robot with a hyperspectral sensor could be watching you from a mile away though, and you'd never know. It could even potentially allow a drone swarm to carry out an ambush of a battalion in waiting, but that's fairly easily solved by never stopping and never assembling in huge morasses except during the attack. Perhaps the birdbot could talk to a cellphone like Nett Warrior and integrate to a battle command network to give you real time updates on a map. That's just Land Warrior at the end of the day though.

Most of the thinking of this stuff was already done in the 90's FWIW, like how in the '50's people thought missile fighters were the future, the problem now is the actual doing. Bill Lind is sort of a lunatic nowadays but he did write a paper or a monograph about this as part of the whole "4GW RMA" dumpster fires that predicted, among other things, mega attacks by hyper terrorists armed with cellphones and sjniper rifles on transformers. I mean, it didn't turn out that way, but it almost did.

Drone swarms were mentioned in one of the AAN powerpoints on DTIC I think. They weren't considered particularly revolutionary. They were way more interested in portable telephonics and cell phones though, precisely because they offered decentralized and highly robust mesh networks that could be used to piggyback terrorist plans on encrypted P2P commo channels.

I'm not aware of any super popular mesh network software because I'm not a protestor or political agitator but AIUI it was pretty important for the Hong Kong protests FWIW.

On the flipside, a publicly available and free to use mesh networking software that allows Android and Apple products to send messages to each other i.e. not bound by cell towers like SMS, would be incredibly useful for post-disaster communications. Sort of a stopgap between ham radio and flying in new cell infrastructure. That sort of exists in USA but it's mostly one way i.e. Amber alerts and tornado warnings and stuff.

 

[zen]

MANeT
Mobile Ad-ho Network

Essentially mobile phone passing data LOS as relays until reaching a tower or ultimately without need of such hard access.

Concept development during the 1990's. Much effort on protocols and code for a self regulating and adapting network.

I read the papers late 1990's researching what could be developed as secure network concepts for LOS by lasers. Keeping mobile armour in the loop through secure comms

 

[sferrin]

There is no small/light weight/inexpensive INS that would allow precise navigation in a GNSS denied environment.

Flies, bees, hummingbirds, pigeons manage it. Nothing special about them that AI won't eventually figure out.

I agree, maybe someday in the future. However, at present such technology does not exist.

The future is coming fast. Just ask all the artists who are starting to realize that the days of actually getting paid to create art are coming to an end, because the AI can create equivalent art that's adequately good, vastly cheaper and vastly faster. AI can fly planes, drive cars, walk robots and hold conversations. Five, ten years down the line, AI capable of flying bird-like drones through bird-like behaviors seem not at all unlikely. Given that there is much more of the future than there was of the past, unless technology grinds to a halt there's no reason to assume that anything not forbidden by physics will fail to come to pass.

This stuff, like designer diseases and designer babies, is coming. It's not just insane but criminally negligent to not start planning for it.

Now combine the artwork with deep fakes of actors and such and we'll have AI making movies with minimal input from humans. "Make me a movie with these actors about a space war with magic robots." Done. Hell, eventually, AI will be saying, "nah, metrics say that's not what you really want. Take a look at this." And it'll be better than Star Wars, Citizen Cane, and The Godfather all put together.

 

[shin_getter]

Tanks is a problem that has been dealt with in the first generation of RMA. Even with the fancy capacities outlined in assault breaker not utilized: hundred and thousands of terminally guided of BLOS munitions directed by long range radar thrown at the problem in minutes, and only very degraded capabilities of old line of sight ATGM, classical high explosive artillery with a tiny sprinkle of guided and even smaller amount of terminally guided long range ammo thrown in, the premier land threat is stopped and the war transitioned to artillery driven positional warfare. People complain about munition costs for missiles, but tanks require professionalized forces and real exercises that is far more expensive than the hull alone, while ATGM works well enough for conscripts and insurgents, never mind manpower minimizing smart missiles with autonomous target ID capacities. Even if new generation of technology enable better anti-tank capability, it is not really battlefield changing, it is like designing a more efficient bullet.

------------------------------
What the focus of drone development has been is anti-concealment: sensor density/fidelity, persistence, low operating cost, especially for militaries intending to attack others: removing the home field advantage is what development is about.

The real potential of "drone swarms" have yet to be realized as many enabling technology is still in the experimental stage. The vision is simple: omni-mobile vehicles that can access terrains and passages inaccessible to anything else, and fight there.

Slaughterbots that breaches walls and go into buildings that kill you, basically. "Drone swarms" kill infantry.

The current practical application of drones have made trenches a increasingly ineffective defense. The next generation is seeking to go into tunnels and buildings. The human with its inability to hibernate with no exposure to the atmosphere and general large size and signature soon can no longer be hidden or evade attack, with firepower is the only defense.
----------------------------
I think any small arms project without considering drone defenses have no future, though small arms in general probably will have relevance of bayonets in <30 years. The future "infantry" will probably there to service the air defense turret on a robot and do high dexterity combat engineering.

 

[Orionblamblam]

Slaughterbots that breaches walls and go into buildings that kill you, basically. "Drone swarms" kill infantry.

Drone swarms can also play an important role in killing armor. Sure, if what you have is an artificial bumblebee with a payload measured in a fraction of a gram, *thousands* of them going "bang" against your tank armor won't do diddly. But you can use them smarter than that:
1) Send swarms of bumblebots loaded with a fractional-gram load of thermite down the 130-mm barrel of the enemy tank, into the barrels of the coax guns and secondary armament. A single such bot *might* set off a high explosive round that's already chambered, might even get into the turret. At the very least they'll trash the armament.
2) Have a single bumblebot land in front of each bit of optics... cameras, periscope windows, laser emitters, etc . Instead of explosives, they are loaded with the better part of 1CC of thinned superglue mixed with very fine aluminum flakes. They either carefully spray, or simply pop, covering the optics with an instantly hard, damned difficult to remove opaque coating. Won't kill the tank, but just *try* to be combat effective when you can't see a damn thing.
3) Fly into the engine inlet and release whatever is most likely to cause a ruckus.
4) If the tank is currently not moving, land thermite bumblebots on the tracks. Cluster and melt down, turning the treads into garbage.

And if you can get close to an enemy airbase, unleash your swarm of bumblebots. Here, they don't need to do anything dramatic, just fly in and land on specific aircraft structures. You know, the bits that are made out of aluminum. And then spray them with gallium, scratch the aluminum surfaces with a fine diamond-tipped claw, sit back and watch the excitement. Extra points for trashing bits of aluminum that wouldn't be immediately noticed until after the aircraft are airborne. If the bumblebots can clamber into the aircraft structure and wait to do their thing until after liftoff, so much the better. Turn the landing gear into mush, for example. If the bumblebots can tinker with the pilot ejection systems to cause the seats to spontaneous blast out of the aircraft without warning, the effects will be *spectacular.* Use tiny thermite charges to turn the pilots oxygen supply into a torch. Burn holes in fuel tanks/lines.


Don't forget that the enemy can do the same thing.

 

[zen]

The smaller the drone and it's onboard systems, the more vulnerable it becomes to various effects.
Such that while small drones deliver tactical reconasense at close range, the probability of being destroyed or rendered inoperable is very high. In fact the smaller the shorter the life.

This makes them short lived, but in that tactical window time critical information can be gained and exploited to advantage.

Swarms are vulnerable to worse than incapacitating attack. A network of sensors building a picture is potentially vulnerable to invasive network attack and doctoring of information. The picture sent back falsified.

A lethal swarm is only lethal if it can exploit a vulnerability. Soldiers in ordinary kit woild be vulnerable to needles carrying toxin.
But such a system is rendered useless against Armoured Soldiers or Armoured vehicles.

It's all going to be a balance of risks, reaction times and capabilities.

 

[shin_getter]

And if you can get close to an enemy airbase, unleash your swarm of bumblebots. Here, they don't need to do anything dramatic, just fly in and land on specific aircraft structures. You know, the bits that are made out of aluminum. And then spray them with gallium, scratch the aluminum surfaces with a fine diamond-tipped claw, sit back and watch the excitement. Extra points for trashing bits of aluminum that wouldn't be immediately noticed until after the aircraft are airborne. If the bumblebots can clamber into the aircraft structure and wait to do their thing until after liftoff, so much the better. Turn the landing gear into mush, for example. If the bumblebots can tinker with the pilot ejection systems to cause the seats to spontaneous blast out of the aircraft without warning, the effects will be *spectacular.* Use tiny thermite charges to turn the pilots oxygen supply into a torch. Burn holes in fuel tanks/lines.

Don't forget that the enemy can do the same thing.

In a tech parity environment, if you have micromachinery to attack, the opponent have micromachinery to defend. The defender have far more mass, energy and support resources to work with, and we know exactly how such things work.

Basically, biology have mastered nanomachinery far before macromachinery. Large scale lifeforms detect, destroy and repair damage by smaller organisms and do so effectively with huge reserves of resources and specialization of the larger lifeform. Small attack vectors can still work, but not by brute force attack but by subverting IFF signals (bacteria) or hacking vulnerable systems (virus).

Once miniaturization enables highly complex micromachinery, it'd go on all the large platforms for various support functions.

The extremely poor energy budget for micro-scale machinery really makes attack difficult in general. You are talking about agonizing over joules of energy. Hardened machinery can be a tough target even when attacked with great precision, and vulnerable systems can simply be sealed within replaceable covers. As such this is unlikely a major threat vector compared to the impact super high sophistication micromachinery would have on everything else.

 

[Orionblamblam]

In a tech parity environment, if you have micromachinery to attack, the opponent have micromachinery to defend.

The Russians and the Ukrainians have tech parity in the field of drones, yet the Ukrainians seem to be using their drones virtually unopposed. Certainly undefended against by *other* drones.

Bug-sized drones might have operational times measured in minutes before the power systems give out. Their entire lifespans might, once activated, be a few hours at most with some battery rechargings. So if you want to defend yourself with them, you need to keep a swarm going non-stop... thousands of new bugs an hour per platoon, perhaps. The attackers, with the exact same restrictions, only need to turn on *their* bugs seconds before they're launched. Even if the bugs are cheap, say, costing the same as a bullet, you're still expending not only money but supply on a constant and ongoing basis. The attacker might only have to send a dozen bugs at a platoon to cause the platoons swarm-defense to exhaust itself for a minute or so... in which time the next wave comes in. A constant low-level harassment campaign of attrition could eat up the defensive bugs.

 

[sferrin]

Slaughterbots that breaches walls and go into buildings that kill you, basically. "Drone swarms" kill infantry.

Drone swarms can also play an important role in killing armor. Sure, if what you have is an artificial bumblebee with a payload measured in a fraction of a gram, *thousands* of them going "bang" against your tank armor won't do diddly. But you can use them smarter than that:
1) Send swarms of bumblebots loaded with a fractional-gram load of thermite down the 130-mm barrel of the enemy tank, into the barrels of the coax guns and secondary armament. A single such bot *might* set off a high explosive round that's already chambered, might even get into the turret. At the very least they'll trash the armament.
2) Have a single bumblebot land in front of each bit of optics... cameras, periscope windows, laser emitters, etc . Instead of explosives, they are loaded with the better part of 1CC of thinned superglue mixed with very fine aluminum flakes. They either carefully spray, or simply pop, covering the optics with an instantly hard, damned difficult to remove opaque coating. Won't kill the tank, but just *try* to be combat effective when you can't see a damn thing.
3) Fly into the engine inlet and release whatever is most likely to cause a ruckus.
4) If the tank is currently not moving, land thermite bumblebots on the tracks. Cluster and melt down, turning the treads into garbage.

And if you can get close to an enemy airbase, unleash your swarm of bumblebots. Here, they don't need to do anything dramatic, just fly in and land on specific aircraft structures. You know, the bits that are made out of aluminum. And then spray them with gallium, scratch the aluminum surfaces with a fine diamond-tipped claw, sit back and watch the excitement. Extra points for trashing bits of aluminum that wouldn't be immediately noticed until after the aircraft are airborne. If the bumblebots can clamber into the aircraft structure and wait to do their thing until after liftoff, so much the better. Turn the landing gear into mush, for example. If the bumblebots can tinker with the pilot ejection systems to cause the seats to spontaneous blast out of the aircraft without warning, the effects will be *spectacular.* Use tiny thermite charges to turn the pilots oxygen supply into a torch. Burn holes in fuel tanks/lines.


Don't forget that the enemy can do the same thing.

The just need to punch a hole in the barrel of the main gun for a mission kill. Wouldn't take much of a shaped charge to do that.

 

[Orionblamblam]

The just need to punch a hole in the barrel of the main gun for a mission kill. Wouldn't take much of a shaped charge to do that.

I wonder if a drone that is good enough to zip down the barrel could get the job done simply by unloading a batch of silicon carbide grit. Probably not, but it might make an interesting test.

 

[jeffb]

https://twitter.com/x/status/1590298403182878722

View: https://twitter.com/TechInsider/status/1590298403182878722?t=8gAUmBqxIScwGI9xSSDXfw&s=19

 

[GruntFox]

In a tech parity environment, if you have micromachinery to attack, the opponent have micromachinery to defend.

The Russians and the Ukrainians have tech parity in the field of drones, yet the Ukrainians seem to be using their drones virtually unopposed. Certainly undefended against by *other* drones.

Yeah, drones are useful against an incompetent opponent. The moment you face someone at least semi-competent, they fail. A lot.

Bug-sized drones might have operational times measured in minutes before the power systems give out. Their entire lifespans might, once activated, be a few hours at most with some battery rechargings. So if you want to defend yourself with them, you need to keep a swarm going non-stop... thousands of new bugs an hour per platoon, perhaps. The attackers, with the exact same restrictions, only need to turn on *their* bugs seconds before they're launched. Even if the bugs are cheap, say, costing the same as a bullet, you're still expending not only money but supply on a constant and ongoing basis. The attacker might only have to send a dozen bugs at a platoon to cause the platoons swarm-defense to exhaust itself for a minute or so... in which time the next wave comes in. A constant low-level harassment campaign of attrition could eat up the defensive bugs.

Or you could do the sensible thing and jam them. The required computing power to make drones semi-autonomous is pretty hefty and impossible to fit on something that small, so they'll need a direct link to a computer.

 

[sferrin]

https://twitter.com/x/status/1590298403182878722

View: https://twitter.com/TechInsider/status/1590298403182878722?t=8gAUmBqxIScwGI9xSSDXfw&s=19

Let's see them catch this:

 

[Orionblamblam]

Or you could do the sensible thing and jam them. The required computing power to make drones semi-autonomous is pretty hefty and impossible to fit on something that small, so they'll need a direct link to a computer.

You would have to provide high-intensity jamming of entire regions 24/7. Your jamming systems would themselves make spectacular targets.

As to the computer power being impossible to pack that small: Miller's Law. Today? impossible. Tomorrow? Difficult. Day after? Available on Amazon. Day after that: in the checkout aisle next to the gum.

 

[jeffb]

https://twitter.com/x/status/1590298403182878722

View: https://twitter.com/TechInsider/status/1590298403182878722?t=8gAUmBqxIScwGI9xSSDXfw&s=19

Let's see them catch this:

Just had an idea for a reboot of "The Jackal".

 

[siegecrossbow]

I don't think anyone raised the possibility of using cheap suicide drones to counter UAVs. They can loiter over the area they defend and when suicide drones approach, they can dispatch them kinetically. This could cut down on the cost and response time considerably since standard guided munition (even MANPAD) cost significantly more than disposable drones.

 

[GTX]

View: https://youtu.be/JRhm0rYFXb4

 

[shin_getter]

In a tech parity environment, if you have micromachinery to attack, the opponent have micromachinery to defend.

The Russians and the Ukrainians have tech parity in the field of drones, yet the Ukrainians seem to be using their drones virtually unopposed. Certainly undefended against by *other* drones.

Bug-sized drones might have operational times measured in minutes before the power systems give out. Their entire lifespans might, once activated, be a few hours at most with some battery rechargings.....

I think a good place to think about the impact of micromachinery is the naval domain.

Ships from 100,000 tons to 20kg saildrones are relevant military assets, spanning multiple orders of magnitudes. It is known that even a massive ship can be disabled by a small precision hit to the propulsion.

So how come naval warfare have no collapsed to mini-vehicles entirely? Here I would introduce a generalized theory of war between machines. On a simple level war is about inducing entropy to a point in space and time, basically delivery of explosions. There are many delivery methods.

Small platforms may be robust, but they are not efficient. Transport ships are 250,000tons as opposed to one man boats, because the former is just vastly more efficient energy and otherwise.

So how do you delivery explosives a long way on top the opponent: the general method is to transport long distances in the "safe rear" with large, vulnerable, efficient platforms. On the front there are inefficient, robust, combat worthy platforms.

So you have a cargo ship move equipment to a forward base, which it goes on a large submarines, which launcher a diver vehicle which places a limpet mine on the opponent ship. Each stage of this action involve a step down in efficiency and a step up in numbers and combat focus.

The same can be said of 100,000ton ships that launches 50ton airplanes that launch 500kg missiles to dispense 50kg submunitions. The larger vehicle increases the efficiency of smaller by such margins that it is worthwhile. Building a force with 40kg warhead missile with 5000km range is just too inefficient even if it have no critical point of failure.
---------------------------------------------
So how does this translate to land warfare? Lets say there is 5km range swarm drones and normally your factories and storage warehouses do not start 5km from the enemy. So you put it on a train, move it 500km from the front, put it on a standard truck, move it 100km from a front, and move it on a military off road vehicle and move it 5km to the front.

But there are other ways to do it. So you can start with a 100km drone, but it is far larger and expensive than 5km range drones so would be outnumbered by 5km drones. As such it is possible to compare different approaches:

If a swarm of 5km drone can protect the launch vehicles from 100km drones, the force can advance to 5km and overwhelm the 100km force complex.

But why drones? The idea is generalized, so rockets, jet engines, launched (gun), one use, multiple use, can fit roughly in the same framework. Can short range X with support platform Y beat long range Z which will be trying to hit the support platform, and under what conditions, and increase analysis to when multiple systems overlap.

So you can compare 5km drone swarm plus support vehicle with 40km artillery with 100km scout drone: can you close?

Note that there are environments and ranges where large platforms simply can not survive efficiently: no man's land. But large platforms can just stay out range and support smaller platforms with scale efficiencies, and only advance after smaller platforms clear out the battlespace.

Now, large efficient mobility platforms are a general rule for long range mobility. Depending on situation, the defense may not need mobility and can forgo the expense and vulnerability of support vehicles and focus on most efficient means. In the naval world this has mean mines, torpedo boats and likes. Well, defense has always had tactical advantages, and this is no different.

 

[GruntFox]

Or you could do the sensible thing and jam them. The required computing power to make drones semi-autonomous is pretty hefty and impossible to fit on something that small, so they'll need a direct link to a computer.

You would have to provide high-intensity jamming of entire regions 24/7. Your jamming systems would themselves make spectacular targets.

As to the computer power being impossible to pack that small: Miller's Law. Today? impossible. Tomorrow? Difficult. Day after? Available on Amazon. Day after that: in the checkout aisle next to the gum.

The biggest problem is the fact that to pump out the required processing power (and note, Moore's Law has the caveat of physics and reality being the upper limits), you'll need a lot of energy, and we've been hitting problems in the solid-state processor field including quantum interference (i.e., we're printing the transistors at such a size that they're literally interfering with each other) and the required cooling dropping efficiency rather quickly. That's why alternative forms of computing are being looked into now because we've hit the limits (or getting pretty close to it) without soft-scifi shenanigans.

To get the semi-autonomous requirements, the computer will burn the poor microchips of an insect-sized bot. So, unless significant breakthroughs in quantum chip manufacturing occur in the next decade or so... yeah, those micro drones are surprisingly easily countered. In addition, all that quantum computing isn't going to be light or small either; as I've said, unless significant breakthroughs occur in the next decade in terms of quantum computer manufacture, we'll be stuck with the massive room-sized units for a chip with maybe a half a dozen q-bits at most (and the only reason why 'classical' computers evolved so quickly is that we didn't hit quantum effects until literally the last decade).

That also doesn't include the possibility of your opponent literally turning your drones against you, as a sat-linked drone got hacked and tricked to force it to land in Iranian-backed insurgent territory. Unless you pull a Horizon Zero Dawn and use literal quantum entanglement for your encryption (and pulling that would likely be problematic as you run the risk of undergoing such a (or similar) scenario with a bad update or QnA didn't catch a few bugs in the software), it will get hacked and turned against you or rendered no more than expensive paperweights.

Note that I know this because I've been working on a few sci-fi settings, and I didn't want to be called out as outlandish by people. Research is a chore for such things.

 

[Orionblamblam]

Or you could do the sensible thing and jam them. The required computing power to make drones semi-autonomous is pretty hefty and impossible to fit on something that small, so they'll need a direct link to a computer.

You would have to provide high-intensity jamming of entire regions 24/7. Your jamming systems would themselves make spectacular targets.

As to the computer power being impossible to pack that small: Miller's Law. Today? impossible. Tomorrow? Difficult. Day after? Available on Amazon. Day after that: in the checkout aisle next to the gum.

The biggest problem is the fact that to pump out the required processing power (and note, Moore's Law has the caveat of physics and reality being the upper limits), you'll need a lot of energy, and we've been hitting problems in the solid-state processor field including quantum interference

You know what's proven to do the job at those scales? Organic brains. Bug brains. Brains genetically modifed and grown in vast numbers in vats specifically for this sort of application seem entirely doable. Genetic modification of bees or flies or cockroaches to serve as the control systems of more capable mechancial drones seem feasible. Grow 'em in situ using drone structures made out of, say, pressed paper. Cheap as dirt, minimal lifespan, biodegradable. Store at low temperature, raise to room temperature in thier transport case, provide specific programming, then release and let the little biomechanical horrors do their things.

 

[GruntFox]

Or you could do the sensible thing and jam them. The required computing power to make drones semi-autonomous is pretty hefty and impossible to fit on something that small, so they'll need a direct link to a computer.

You would have to provide high-intensity jamming of entire regions 24/7. Your jamming systems would themselves make spectacular targets.

As to the computer power being impossible to pack that small: Miller's Law. Today? impossible. Tomorrow? Difficult. Day after? Available on Amazon. Day after that: in the checkout aisle next to the gum.

The biggest problem is the fact that to pump out the required processing power (and note, Moore's Law has the caveat of physics and reality being the upper limits), you'll need a lot of energy, and we've been hitting problems in the solid-state processor field including quantum interference

You know what's proven to do the job at those scales? Organic brains. Bug brains. Brains genetically modifed and grown in vast numbers in vats specifically for this sort of application seem entirely doable. Genetic modification of bees or flies or cockroaches to serve as the control systems of more capable mechancial drones seem feasible. Grow 'em in situ using drone structures made out of, say, pressed paper. Cheap as dirt, minimal lifespan, biodegradable. Store at low temperature, raise to room temperature in thier transport case, provide specific programming, then release and let the little biomechanical horrors do their things.

Here's the thing, we don't know how to program brains at all. Even if you could make factories for cybernetic bug brains, we don't know where to begin in terms of programming them. Hell, we're still scratching our heads about how brains do things more often than not.

 

[shin_getter]

You know what's proven to do the job at those scales? Organic brains. Bug brains. Brains genetically modifed and grown in vast numbers in vats specifically for this sort of application seem entirely doable.

It is trivial to make a nation with the capabilities of Mr. Von Neumann (what, dig up a grave, cross some genebank data and spend some time CRISPR people: all doable today if ethics boards are killed) compared to making bugs do combat missions. Basically it'd be something that happen on the other side of transhuman (intelligence explosion: singularity) horizon and I don't really think about that.

The biggest problem is the fact that to pump out the required processing power (and note, Moore's Law has the caveat of physics and reality being the upper limits), you'll need a lot of energy, and we've been hitting problems in the solid-state processor field including quantum interference without soft-scifi shenanigans.

Depending on mission complexity, not a lot of processing power is necessarily needed. Or more correctly, systems would just be designed with the available processing and expected communications available.

Reaching limits with current processing paradigm. However there are materials like graphene and concepts like spintronics still undeveloped. There is orders of magnitude possible increase in efficiency with different model of computation: it takes millions of dollars in power bill to train something that a human largely learns off a 12watt brain. Even something as smart as a wasp can be a problem.

The future is long time, so what can be developed probably will happen eventually.

 

[Orionblamblam]

Here's the thing, we don't know how to program brains at all.

And how long do you think that state of affairs will last? Willing to bet that some government somewhere isn't right now plugging bug brains into AI systems and setting the AI the task of figuring them out?

But the point remains: small bits of goo are able to do what computers are said to not be able to do. Unless you are willing to suspend reason and assume that living brains have some supernatural element to them that makes them capable in ways artificial systems never will be... we will, in time, be able to make artificial thinkin' machines that are on the scale, capability and power requirements of organics. And then they'll do *better.* No reason to assume that a rat brain won't fit on a chip. Overclock the thing, attach it to a teeny camera and some fins, and now you have an angry, vengeful bullet that will steer itself towards the target that was programmed into it by the optics on the gun in the moment it was fired. Sure, the chip might burn out in three seconds... but it'll be done by then anyway. A dandy anti-drone device, perhaps.

 

[Kat Tsun]

The PLA is probably using Falun Gong brains but I doubt they're having very much success. We're also reaching peak transistors, so computers will likely stagnate over the coming decades, like how internal combustion engines did in the 80's. I suspect a decade from now GANs will still be producing incoherent semi-dreamlike ramblings rather than coherent sentences. Frightening for horoscope writers but not much else.

 

[GruntFox]

Here's the thing, we don't know how to program brains at all.

And how long do you think that state of affairs will last? Willing to bet that some government somewhere isn't right now plugging bug brains into AI systems and setting the AI the task of figuring them out?

But the point remains: small bits of goo are able to do what computers are said to not be able to do. Unless you are willing to suspend reason and assume that living brains have some supernatural element to them that makes them capable in ways artificial systems never will be... we will, in time, be able to make artificial thinkin' machines that are on the scale, capability and power requirements of organics. And then they'll do *better.* No reason to assume that a rat brain won't fit on a chip. Overclock the thing, attach it to a teeny camera and some fins, and now you have an angry, vengeful bullet that will steer itself towards the target that was programmed into it by the optics on the gun in the moment it was fired. Sure, the chip might burn out in three seconds... but it'll be done by then anyway. A dandy anti-drone device, perhaps.

Given that we're still trying to figure brains out, even now? Likely something on the timescale of multiple decades, minimum.

In addition, you're not taking physics, quantum mechanics, and reality into account here. We're already hitting the top end of what solid-state electronics can do, and the most likely prospect for AGIs is quantum computing, which is barely in its infancy right now (the reason that solid-state electronics evolved so fast and so far is that we didn't have to work with quantum weirdness right out of the gate).

There are two reasons why we didn't go bio-computer: 1) we didn't (and, even now, still really don't) have the tech to even go for this sort of path, and 2) we don't know a lot about how brains function (and now, the intervening steps between A and up to K).

The biggest problem is the fact that to pump out the required processing power (and note, Moore's Law has the caveat of physics and reality being the upper limits), you'll need a lot of energy, and we've been hitting problems in the solid-state processor field including quantum interference without soft-scifi shenanigans.

Depending on mission complexity, not a lot of processing power is necessarily needed. Or more correctly, systems would just be designed with the available processing and expected communications available.

Reaching limits with current processing paradigm. However there are materials like graphene and concepts like spintronics still undeveloped. There is orders of magnitude possible increase in efficiency with different model of computation: it takes millions of dollars in power bill to train something that a human largely learns off a 12watt brain. Even something as smart as a wasp can be a problem.

The future is long time, so what can be developed probably will happen eventually.

Not when you consider the likely cyberattacks to make them inert or subverted. We've already had a top-of-the-line US stealth Sat-Linked drone (i.e., what can be considered unhackable outside of extraordinary circumstances before this) be tricked by a bunch of insurgents with off-the-shelf equipment and some pointers by a nation-state to land in Iran/Iranian-backed Insurgent territory. Given that the drone paradigm favors one who can get the most drones out and utilize them, it is likely that nation-states like Iran or North Korea are working on software packages that render any that doesn't have an AGI fitted either useless or worse than useless.

Everyone in this thread constantly forgets that there are multiple points of contact in drone warfare, and the enemy gets to vote in those areas too.

 

[shin_getter]

We're also reaching peak transistors, so computers will likely stagnate over the coming decades, like how internal combustion engines did in the 80's. I suspect a decade from now GANs will still be producing incoherent semi-dreamlike ramblings rather than coherent sentences. Frightening for horoscope writers but not much else.

In addition, you're not taking physics, quantum mechanics, and reality into account here. We're already hitting the top end of what solid-state electronics can do, and the most likely prospect for AGIs is quantum computing, which is barely in its infancy right now

There is still orders of magnitudes speed up possible by converting hardware from general purpose to single purpose even without new scientific development, just more engineering work. Historically investment in new general purpose hardware was more valuable and this branch is underdeveloped.

This underdeveloped branch will enable computing to improve for decades from this point on, if at worst engineering effort to results improvement than the previous case.

For the general problem of AI, algorithmic improvement is still happening at a steady pace, it works like a search problem. The solution space is huge and is unlikely we've reached anywhere near optimality. The huge amount of compute require is due to needing to brute force solutions: once discovered things like AI models can be compressed massively.

The shock of GPT and is how a text token prediction model without any capability for recursion is capable of "near intelligence" by brute forcing it. It is not an system architecture with provisions for "higher intellect."

Some people think the brain is amazing, but ultimately it has huge switching elements running at single digit Hertz, compared to quantum scale switching elements running at gigahertz. The algorithmic efficiency is likely the magic sauce. High temperature quantum computing is unlikely.

Not when you consider the likely cyberattacks to make them inert or subverted. We've already had a top-of-the-line US stealth Sat-Linked drone (i.e., what can be considered unhackable outside of extraordinary circumstances before this) be tricked by a bunch of insurgents with off-the-shelf equipment and some pointers by a nation-state to land in Iran/Iranian-backed Insurgent territory. Given that the drone paradigm favors one who can get the most drones out and utilize them, it is likely that nation-states like Iran or North Korea are working on software packages that render any that doesn't have an AGI fitted either useless or worse than useless.

Everyone in this thread constantly forgets that there are multiple points of contact in drone warfare, and the enemy gets to vote in those areas too.

The world has seen effective autonomous weapons long before communications. Consider the landmine.

The world has seen effective use of unmanned aircraft without communication. There was the cruise missile. There was photorecon. There was autonomous target seeking systems like Brimstone and Harpy, already in service for decades. Then there is recon by explosion: the absence of enemy action suggests security.

The world has attempted to defeat radio communications ever since it was invented. Decades of attempt have not produced a reliable general solution.

------------
The problem of jamming is almost self solving: a jammer is the inverse of stealth and is a easy target for autonomous detection and attack. Anti-radiation loitering munitions was one of the first application of "drones." Of modern states, Russia was the better supplied forces with electronics warfare equipment, and in open conflict its applicability is found to be limited. Broadband jamming consume huge amounts of fuel, jams friendly communication and is a dead obvious target. Targeted jamming on specific protocols avoids some of these problems, but does not work if the opponent changes their communication details, and still can be localized with electronics warfare equipment and shot at.

One thing that can be said to be surprising in this conflict is that successful widespread use of commercial non-jam resistant drones. Jamming is only a conceivable solution for forces with fire superiority, and even than only against vehicles with specific missions.

Computer target recognition is more than good enough for a lot of roles. It really depends on the target and mission in question. For something as obvious as a ship, even ww2 sensors suffice. If highly concealed fixed positions, even humans can fail identification at close range.

To proclaim that drones are OVERRRRRR because one is proclaimed to have been jammed once is like saying Airplanes are OVERRR because someone with a rifle shot one down once!

The decades long struggle of electromagentic warfare can not be summed up with "lol jammers win." I guess in the case where jammers do win, communication would go back to pigeons and drones that mimic pigeons in function.

And then there is wired drones, as 10km recon drones was thought as a good idea in the era of fiber optics guided munitons.

Then there is optical laser communications, where jamming is impossible..... etc etc

Electronic warfare imposes costs and friction on the opponent, it has never ended communication altogether.

 

[1635yankee]

I keep coming back to a cellphone.. Your average cheap phone can track a face just fine. All it needs is to be programmed to recognise the contrast between sky and flying dark thing. Also, a lot of phones now have multiple cameras.. Different angles of view and spectra maybe? More than enough processing power to turn that into direction information.

Suggestion: take your phone outside, set the camera to the wide lens, photograph a sparrow or a hummingbird flying a block or two away, see what it looks like. How many pixels wide is it?

hence aquiring with the narrower lens and tracking with the wider?

Any antidrone weapons would need to cheap and small, like drones, so that a they can be used on the scads of drones that can appear. This is very challenging! What may be the ultimate solution is "fighter" drones armed with a submachine gun. They would not be that easy to develop, and autonomous weapons scare the crap out of me (Skynet, anyone?)

Spoofing these will be a problem, as engineering constraints will tend to make all drones alike. The chances of hacking can be minimized by not making them programmable in the field. Of course, you'll still have to worry about people being corrupted and hacking farther up the production and service chains (like the company that had its hard drives pre-installed with a virus/trap door from the factory a few years back or the backdoor the US government wanted on all computers at one time).

 

[Orionblamblam]

There are two reasons why we didn't go bio-computer: 1) we didn't (and, even now, still really don't) have the tech to even go for this sort of path, and 2) we don't know a lot about how brains function (and now, the intervening steps between A and up to K).

How long did it take for society go from "genetic tinkering with humans is sci-fi nonsense" to "did you see Billy Joe Bob's latest genetic biohack to himself?" Seems to me it was about roughly the same length of time it took to go from "video phones are just too difficult and clumsy" to "I keep it and the supercomputer it works on in my pocket and use these technologies that superpowers would have gone to war over in my parents day in order to make TikTok dance challenge videos."

Everyone in this thread constantly forgets that there are multiple points of contact in drone warfare, ...

Everyone? That includes you.

 

[1635yankee]

There is no small/light weight/inexpensive INS that would allow precise navigation in a GNSS denied environment.

Yet. With onboard cameras, how much accuracy needed by the INS may not be that great, perhaps tens of meters after a few kilometers, not the sort of accuracy needed by the INS in aircraft or SSBMs.

As an aside, I can order an INS for the robot built by my students for First Robotics for about $30.

 

[TomcatViP]

IMOHO, the revolution is not in electronics, sensors or weapons.
It's in thinking.
When you see that idiots offering something as cumbersome and pricy as a a net catching drone, a laser frying everything and the attending public or a system based on something as heavy in manpower as trained birds of prey that needs trainers, regularly keep getting public funds.. well many engineers interested in solving this problem must have rest seated thinking to themselves: but what for?

 

[zen]

Jamming....
The reason the Russians are not letting rip with their extensive EW kit is a simple one of actually needing it themselves. A feature of systemic corruption and incompetence. Hence troops relying on civil mobile phones, GPS and all the attendant mayhem we've seen.

Had they got their thinking straight, they'd have shut down civil comms and gone 'old school' on Communications. Which however bad it might seem is hard to penetrate.

The alternative is to flood channels with rubbish data and hide amongst it.

 

[TomcatViP]

Bandwidth will prevent this strategy from being successful. It can work for hackers but not for military that need a large flow of data.

 

[zen]

Bandwidth will prevent this strategy from being successful. It can work for hackers but not for military that need a large flow of data.

I'm not sure the military need 4K TikTok videos?

 

[TomcatViP]

IMOHO, Military data transfers on open networks, by being extensive, out of common place and use, should be traceable.

 

[GruntFox]

There are two reasons why we didn't go bio-computer: 1) we didn't (and, even now, still really don't) have the tech to even go for this sort of path, and 2) we don't know a lot about how brains function (and now, the intervening steps between A and up to K).

How long did it take for society go from "genetic tinkering with humans is sci-fi nonsense" to "did you see Billy Joe Bob's latest genetic biohack to himself?" Seems to me it was about roughly the same length of time it took to go from "video phones are just too difficult and clumsy" to "I keep it and the supercomputer it works on in my pocket and use these technologies that superpowers would have gone to war over in my parents day in order to make TikTok dance challenge videos."

... that is ignoring the difficulties of different fields. Cameras and classical computers as a whole didn't have the quantum weirdness that dominates quantum computing or the sometimes bizarre world of genetic sequencing interactions.

While biotech has gone a long way since the 1980s/1990s, it's far harder than it initially looks despite us offloading the real number-crunching to computers.

Everyone in this thread constantly forgets that there are multiple points of contact in drone warfare, ...

Everyone? That includes you.

No, I see the other points of contact and the interactions with drones.

 

[Orionblamblam]

Everyone in this thread constantly forgets that there are multiple points of contact in drone warfare, ...

Everyone? That includes you.

No, I see the other points of contact and the interactions with drones.

Wait. If it's "everyone" that includes you. But if it doesn't include you, then it's not "everyone." You've introduced a recursive paradox that will destroy the universe! You maniac!