dan/california

Interesting video of a guy trying to break the drone speed record. It can be long and slow in places, so I will summarize. A moderately competent two person father and son team with a nice garage shop build a drone that can go 500KPH. They have about .000000001 percent of the resources of Lockheed-Martin. This is exhibit A of we haven't seen anything yet, and BTW helicopters are OVER. 

I am not sure those are considered a negative factor anymore.  In fact those just might be the folks they want now....

https://mastodon.social/@ChrisO_wiki/112472540071030065
Kind of to be expected that especially members of security / police etc. Are going to the war in Ukraine and dying there. These are missing inside Russia now. 

Above is one of the clearer examples from WW2, bad weather and bad choices had bad results, the war went on. The bombers came back the next day with a better visibility and a better plan. The breakout that resulted pretty much went all the way to Rhine.
In the current conflict at least 5%, and perhaps a multiple of that, of the Russian Air Force's casualties have been friendly fire. I am sure they aren't happy about it, but they haven't had the sense to quit and go home, yet at least. Even the Ukrainians have lost a few planes to friendly fire. Brave men get in the cockpit every day and go out again anyway.

First of all there are no truly reliable methods of ensuring civilian safety in an active combat zone, there never have been. There are armies and nations that try to be discriminate, and armies and nations that don't bother. But perfection in a peer to peer shooting war where both sides are jamming and obscuring everything they possibly can, in any way they can think of is unattainable. The expression that there is no such thing as friendly artillery is about as old as anything that deserves to be called artillery.
The thing we aren't talking about enough is that there are degrees of jamming, and degrees of control. A live video going one way, and second by second flight commands going the other are by far the situation with the highest requirements for consistency and fidelity. As soon as anything less than that is required all sorts of intermittent communications options become viable. A the very simplest level you send out a first wave of drones to engage the very front line of enemy positions. You simply tell that first wave to self destruct or take whatever its current best target is, or simply hit a set of coordinates that was thought to be relevant at a specified time. So now, even in a 100% jamming environment your other forces are clear to proceed to their first phase line. At the next level up you have expendable transmitters that can broadcast a short but very high power signal to tell your drones to change kill boxes, and can even have a safety that any drone that doesn't receive a signal by the time specified crashes /self destructs. And the variations go on forever, but all of them are vastly more tolerant of reduced bandwidth than needing full time video one way, and full time control signals the other.

Some time ago I spent a bunch of time at work talking to one of the metrology guys about photogrammetry, which at the time took a special camera and software, coded stickers, calibrated sticks, and the like, and cost about $70K for the hardware.  I think about 5 years later you could get pretty decent performance from a $1000 iphone without any of the extra stuff.

What happens, if you set tank speed as "fast" for minecleaning with a tank plow. Russian "barn-tank" likley retreated on full speed and its plow just drove over the mines too fast, so mines blew up under the tracks. It's amazing one crewman has survived
 

The solution to this 'billboard problem' is to fly a half circle around each target as standard practice. You might want to do that anyway to get a better attack angle.
Additionally, you train the model with pictures of tanks on billboards.
Btw, how long do you think would a billboard last on the battlefield?

There is definitely software that can build an accurate 3D model of an area based on data gathered from a moving camera. That is not a capability that an image-recognition AI would normally have. It isn't hard to image pairing the 3D modeling capability with the image recognition capability. But remember the first rule of AI. What's hard is easy, and what's easy is hard.
You'd need extra logic to prevent it from spitting out an answer other than "yep, there's a tank at the location of that flat rectangle in the 3D model". You might solve that by building a 3D model recognition AI. But it probably won't always be possible or practical to build a full 3D model, so you'll want to retain the flat image recognition software. But there will be cases where the image recognition software will say an object isn't a tank, while the 3D model software will say that it is a tank and visa-versa. Which do you train the overall program to trust?
For now I think the answer is manned-unmanned teaming (swarms of unmanned platforms controlled from a single manned command platform). If what's hard for humans is easy for AI, and what's easy for humans is hard for AI, then having them work together just makes sense (at least until we can figure out how to make what's easy for humans easy for AI). I think manned-unmanned teaming is the route we're going down in the near-term. It allows us to use AI on the battlefield right now, without having to wait for it to get more advanced, while effectively compensating for the weaknesses of modern AI. And it seems like a natural mid point between a fully manned force and a fully unmanned force (it seems plainly obvious to me that fully unmanned is the way things will be done in the far future).
For sure our 6th gen fighter program is going for manned-unmanned teaming. But I think ground warfare is going to go down that route as well sooner or later. Imagine a tank platoon, or possibly even a tank company, in which all of the actual "tanks" are UGVs and the only manned platform is the HQ vehicle. Or an artillery battery in which all of the guns are UGVs, with their fire directed from a single HQ vehicle. I could even imagine a mortar platoon in which all of the tubes are mounted on UGVs controlled remotely by a specialist from the company HQ.

This is a long, fairly technical article about how apple puts together a three D scan of a room. A decent autonomous drone control system would a mostly analogous process.
And as Steve has been emphasizing, in a real war perfection is not a requirement.

I heard from American acquintances who'd served in Afghanistan at least ten years ago about multiple instances of senior NCOs who effectively hijacked mast/balloon mounted surveillance cameras to spy on their own patrols to ensure that no one was violating petty uniform regulations.
"Tell Private So-and-so to get his eye pro back on. And by God Sergeant, roll down your sleeves again!"

Not sure this one has surfaced here yet:
Obviously there's a lot of focus on the positives of drones, but how often do we think about how much they can encourage higher commanders to micromanage?

A moving camera has 3D information.
I used to bike race with a guy who had only one eye.  He could not only ride close in a paceline at speed just fine, he could yell at the person in front of him for not following close enough.
Training the algorithm is the part that requires a lot of computing power and data. You can run it on something much lighter weight than you use for training.  A rough example (I don't know that they even use ML for it) is terrain relative navigation for Mars landings.  That runs on a machine that's got the capability of the middle of the line 1998 mac laptop (Rad750 at something like 200 MHz, not even as fast as the "high end" WallStreet).  That machine doesn't do all the pre-processing necessary to make it possible, it just takes video input and drives actuators.  
(ETA: I did a little search and the Mars helicopter does all its navigation using autonomous feature detection in real time.  It's running on a snapdragon, which is the real significance of the helicopter - it's many generations later than the computer that runs the rover that drives it around.  The light-time to Mars makes joysticking impossible.)

My point was that your point was silly and unrealistic.  But if you have forgetting my point then yours has no doubt retreated in the darkness never to return.
An entire military trade built around error and missing - not an entirely surprising outcome.

This part of today's ISW update made me chuckle.
Can you shoot an AI for cowardice? I would think they would find a way.
Armando Iannuci needs to get his pencil out when all this is over.

Yeah, like you’ve been saying so they push and take some territory with big losses, but then what? Unless they can push on a broad front, they won’t be able to sustain their push, and it’s not clear they forces capable of this. This isn’t a pocket they are trying to close.
So what’s next? The arrival of F16s now that the air defenses (and radars in particular) in Crimea have been degraded?

"Modern" attack helicopters are completely obsolete. People have just spent so much money on them over the years that don't want to admit it just yet. Between cheap, expendable, and VERY soon autonomous drones, and things like the later models the Israeli spike missile that have a ~40km range, it makes zero sense to invest in expensive helicopters. At least the U.S. canceled the program for a next generation model.
The Russians got good use out the Ka-52s last year because of all the things NATO would bring to this fight that Ukraine didn't have. Good airborne radar at the right range is just one of MANY of those capabilities. They were also using the helicopters because they were they only thing in inventory that could fire their very best long range ATGM. I don't think anyone else is going to make that mistake again for a while. Whatever you want to say about the helicopter it is a very good missile. Ukarine is lucky tey don't more of them on more platforms.
These are smart motivated people, but at the same time this third year engineering student stuff. They are throwing large amounts of semiskilled labor at the problem to get thee production they need. For Ukraine, at this moment, this necessary and appropriate.
So a lot of discussion about drone costs. A new Iphone costs less than $2000, by even the most pessimistic assumptions. It has an approximate infinity of processing power, and three great cameras. There is just no reason for the brains of a drone to ever cost more than that. so even if all the other bits, including a nice tandem/EFP warhead come out to $5000, you still have a DELUXE FPV drone for $7000. Except it won't be FPV, all the operator will have to do is confirm the coordinates of the kill box, and pull the safety on the warhead. 
The Pentagon needs to invest in the drone equivalent of a Gigafactory to make them by the tens and tens of thousands. And  they need to have a come to Jesus conversation with the defense industry about the way they get paid. We can afford to overpay somewhat for hardware engineering and development that works. Getting overcharged on a per piece basis is just not viable anymore.

Well, given that drones are changing warfare all by themselves, that seems like the least they could do, no?
Even if they're not going to make it, they could at minimum deliver it.

First of all there are no truly reliable methods of ensuring civilian safety in an active combat zone, there never have been. There are armies and nations that try to be discriminate, and armies and nations that don't bother. But perfection in a peer to peer shooting war where both sides are jamming and obscuring everything they possibly can, in any way they can think of is unattainable. The expression that there is no such thing as friendly artillery is about as old as anything that deserves to be called artillery.
The thing we aren't talking about enough is that there are degrees of jamming, and degrees of control. A live video going one way, and second by second flight commands going the other are by far the situation with the highest requirements for consistency and fidelity. As soon as anything less than that is required all sorts of intermittent communications options become viable. A the very simplest level you send out a first wave of drones to engage the very front line of enemy positions. You simply tell that first wave to self destruct or take whatever its current best target is, or simply hit a set of coordinates that was thought to be relevant at a specified time. So now, even in a 100% jamming environment your other forces are clear to proceed to their first phase line. At the next level up you have expendable transmitters that can broadcast a short but very high power signal to tell your drones to change kill boxes, and can even have a safety that any drone that doesn't receive a signal by the time specified crashes /self destructs. And the variations go on forever, but all of them are vastly more tolerant of reduced bandwidth than needing full time video one way, and full time control signals the other.

You want it to make you OJ in the morning too? Human beings do fratricide all the time under those conditions. In fact AI will likely do better because it’s processor is not built on a panicky teenager chassis and when told not to shoot, it will actually follow orders.  If you wanna set the bar that high we may as well outlaw all warfare.
 

Interesting video of a guy trying to break the drone speed record. It can be long and slow in places, so I will summarize. A moderately competent two person father and son team with a nice garage shop build a drone that can go 500KPH. They have about .000000001 percent of the resources of Lockheed-Martin. This is exhibit A of we haven't seen anything yet, and BTW helicopters are OVER. 

I said president of Russia (in rus. genetive Rossii), but not Raisi ! 

Yeah, that’s the real question: “How smart do you need it to be?”
Recognizing useful targets, navigating using visual references, cooperating with other drones to attack a group of targets, attacking weak points of targets… that all seems within the whole budget side of things.
For example, a model that is accurate for identifying all extant armored vehicles and their weak points (including rolling garden sheds), that’s just not that big of a model. Fits on a phone easy peasy.

Totally true.  In reality this little beast is kinda shooting for the deep end - asymmetric warfare and HVT strikes. This is in the Switchblade 300 league - not go out and find me a tank and kill it with 2-3 buddies.  But the video does show where I think things will go in the short (read “now”), the last mile.  If an operator can get the drone within a few kms, do the target selection, and then hit release, we have a drone that is basically a flying ATGM system.  It can’t be soft-killed on final approach as there is no link to operator.
Second, these videos are from 2020. I am fairly confident that processing power has continued to increase and what can be built now for less has higher capabilities.  Regardless the age of full autonomy is going to happen, it is already happening.
The main issue with EW is that it takes a lot of energy to get what you want, in an age where pumping a lot of energy out is essentially self-illuminating targeting.  Direct energy weapons will have the same problem. Large, big and brash is not going to fair well in an age of small,  cheap, everywhere.  It would be better to figure out how to create distributed EW capability and put them on unmanned systems as a net as opposed to single major platforms.

I am so tired of this entire line of thought.  Sure the UA has legacy Soviet doctrine but 1) what is wrong with Soviet doctrine - there is a huge assumption that Soviet doctrine was ineffective and somehow did not know how to manoeuvre, which is patently false, and 2) And ignores the other factors staring them in the face that are making large scale manoeuvre impossible.
This is straight up western arrogance built on a foundation of ignorance.  Until someone from the Pentagon can explain how we would successfully operate under the same conditions (no air superiority, ISR every where) using “mission command and manoeuvre” of course.  Did anyone ask the Ukrainians why they are using fires first?  Did anyone stop to wonder why manoeuvre has failed for the Russians as well?  Nope, too easy to blame it all on “Soviet-era” doctrine.  This “us-centric” analysis is frighteningly similar to what European observers of the US Civil War walked away with…and it cost them dearly.