chrisl
-
Posts
2,102 -
Joined
-
Last visited
-
Days Won
2
Posts posted by chrisl
-
-
38 minutes ago, dan/california said:
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.
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.
-
1 minute ago, Butschi said:
How so? Sorry but your statement is really easy to falsify. Just use my billboard example: If the camera moves straight towards a 2D plain that is perpendicular to the movement that camera has exactly zero 3D information and no way of finding out that it is a 2D plain.
I know what you intend to say, you can derive 3D information, making assumptions, using heuristics and in the right circumstances. If you see a car moving you can derive distance and speed once you make an assumption about its size and shape. You use your experience to do that (a child will have a harder time than you have) but you can be entirely wrong if the car you assumed to be 5m long is, in fact, just a toy car of 5cm.
Moreover that isn't how simple detection and tracking works. Usually detection is frame by frame and afterwards tracking tries to match the indivual detections to each other (maybe priming detection to take a closer look at a region of interest). But it is not "ok we now have 100 detections of this tank, lets put them all together and see if it is really a tank". You can always do better, sure, but better is more expensive.
This isn't necessarily true, either. Depends on how the training is done and how inference is done later. For instance, in some cases, when using the network afterwards you draw many samples from some distribution. During training often just one sample was used (but for a batch of different input data) whereas later on you draw many samples for just a single input situation. That is not necessarily less computing (or memory!) intensive than training. Also, during training you don't care about things like real time capabilities. Coming back to our drone, it can really make a difference if you can make updates with 0.5 Hz or 10 Hz. Even if training needs more compute power that doesn't mean you can run every trained model in real time on a phone.
So if the camera moves at exactly the right (or really wrong) path, you can't see that the thing is flat. So what. Drones aren't going to just be moving in a straight line and not gimbaling the camera. You're thinking too automotively. You don't even need heuristics or assumptions if you have a reasonable IRU (which can be done on a chip) - you can fly around and build a 3D model. My dentist does that realtime now with a stick that they wiggle around in my mouth to make crowns, instead of making a casting.
Tracking without doing individual frame by frame detection is not only possible, but there are multiple ways to do it. Doing frame by frame and stitching sucks when you have low SNR or are near the resolution limit, but you can improve SNR by taking advantage of relative motion of the camera and scene.
-
52 minutes ago, Butschi said:
I was thinking about a much simpler method. Have a few cardboard billboards with images of tanks on them placed prominently. A simple image recognition network that has only a single camera as sensor has no 3D information and so has no way of deciding if the tank is a real tank or just a flat image if it comes more or less at a right angle. Place the others covering different angles. Sure, a stationary drone trap but a very cheap one. Mount them on cars for moving targets.
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.)
-
3 hours ago, Butschi said:
To be fair, that video is really on the Puma vs drone swarm video level we were discussing before. It demonstrates a swarm flying in a more or less static formation at conveniently high altitude. That doesn't really qualify as AI in my book as it should be fairly easy to do with a few sensors and conventional simple algorithms. The Wikipedia page explicitly says: "The capability of this swarm to autonomously identify, select and coordinate attacks on a target has, however, never been demonstrated by STM in reality."
Re: the discussion how expensive autonomous UAVs would be: It depends, I guess. As others have posted already, there is a lot that can be done with mobile phone hardware which isn't really expensive. In general, though, AI is such a vague concept these days that it really depends on exactly what we are talking about. For some stuff a mobile phone is sufficient. Many things require a decent GPU, though - so, a good gaming PC - and others (think ChatGPT) are more on a computing center level.
That said, once you have trained your network you can often either put it on a dedicated ASIC or use it to train a much smaller network that can run on cheaper hardware.
So... at the end of the day it boils down to "how smart to do you want it to be?". I guess(!) that patrolling an area at above tree top level, applying some image recognition and tracking and then going straight for that target is really not beyond what a mobile phone can do (and that is probably something that is already done). That Wikipedia page says something about face recognition which is certainly something my phone can do to unlock the screen. Then again, it clearly says that it is possible that others with a similar face may be able to do that, too. And on a Monday morning before I had my first coffee, it has a hard time recognizing even me.
And that is probably the gist of it. The more you invest in hardware, the better your result is going to be. If you are only interested in your drone killing something that looks like the typical tank out in the open and don't care if it hits a civilian car every now and than, then, that is certainly something a phone can do. If you want to hit one guy in crowd with absolute certainty and noone else... maybe not.
You're overthinking the autonomy and mentally turning "autonomy" into "AGI".
A few sensors and some rules is probably sufficient if you can send them to an environment where there are no friendlies. The rules can be deterministic linearly programmed (automated) or they can be developed via some ML algorithm that gives results that may be less obviously explainable, but they work, and need to still be bounded by some deterministic rules ("don't let the autonomy run unless you're in this region"). I can make an autonomous system that will process video data and run on a 5 year old snapdragon. It's not going to have AGI or be able to write sloppy term papers, but it will identify and prioritize objects of interest, even if it hasn't seen them before.
-
45 minutes ago, kimbosbread said:
Why prohibitively expensive? A modern phone has all the processing power you need; equivalent to laptop basically. These chips aren’t expensive and are produced at massive scale. Autonomy and machine vision don’t mean general AI, they mean the ability to tell a drone “go over here, and find targets that match X, Y, Z characteristics”.Yeah, think of the drones as replacing Javelins and mortars, except with the range of artillery. You are basically pushing mini cruise missiles down almost to the squad level. If a lot of squads have these, and coordinate, you can in fact get density of fires.
For recon, why does a comm link need to be expensive? If we can do starlink to phone, then we can do starlink to drone. If there’s too much jamming, the designated anti-radiation drone can go find the jammer (or a bigger munition can be dedicated to that). And the jammers don’t seem to have that great range, especially the more portable ones. So you just head up a bit higher, no problem. I don’t see this as a deal breaker at all.
Sure, but likely it’ll operate more like a submarine where the command channel is mostly listening. Moreover, it will be along a wide frequency band, and this will force the jammer to be much noiser and more visible.
And if it's doing recon and you let it be autonomous, or even just automated (pre-determined flight path), jamming only affects the reception, anyway, it doesn't prevent a drone from transmitting. And if there's nothing jamming the receive station, there's still realtime recon.
A jammer would have to have enough power (and maybe LOS) to get to the receive station to stop that. A local jammer might affect a nearby relay, but if the relay is some distance away it's likely to be unaffected, or significantly less affected. If the EW is powerful enough to cause interference all the way back to the operator, the next drones that launch will have the RF homing module on them.
The most effective EW would be triangulation systems to detect drones' transmissions and then shoot them down, either the old fashioned way or with another drone.
-
3 minutes ago, Letter from Prague said:
Addendum: "breaking" modern encryption isn't really a thing. You would need quantum computers that don't really exist at the moment, or more time than until the end of the universe to do something like that - even the dumbest CPUs that cost $1 a piece have integrated modern encryption nowadays.
What is usually attacked is when whoever built the device, hardware or software, made a mistake or cut some corners, like not implementing encryption at all or not generating encryption key properly or messing up some code or even left a backdoor. That's where you might see a difference between Raytheon and DJI drones.
I left open the possibility of "breaking" because good encryption costs energy, and drones are on a tight energy budget. If you can get even a little more range or patrol time in exchange for something that can be decrypted in a week, it's probably worth the trade.
-
29 minutes ago, poesel said:
I haven't seen that tech in this war, either. But you dismissed EW for the next war, and I think this is too early to do.
Even if a drone is autonomous, you still might want to have a command channel to it, I refer to Dr. Strangelove for the mechanics of it, although the encryption mechanics have improved a bit since then.
But even a fully autonomous, no back-channel drone can still be fried by microwaves or blinded by a laser.But I have to concede the point that the importance of the necessity of a continuous radio connection to a drone will go down drastically. Hence, old-fashioned EW will lose its attack vector and become much less useful.
To answer the question of why not just hack the entire network? Because encryption on anything from mobile phone up is so hard that you need really, really much computer power to break it. OTOH, support for encryption on microcontrollers that usually fly those drones has been lacking. It just costs too many CPU cycles to do good encryption, and don't think about the video stream until you have specialized hardware.
But this, too, will go away in a few years, of course. The demand is there, so there will be products.I suspect there will end up being a big asymmetry in EW capabilities. The current Russian approach sounds mostly like "spray and pray" - send a lot of signal up in the relevant bands to cause interference, but it's not always effective because even consumer stuff (maybe even especially consumer stuff) is designed to accept large amounts of interference because of FCC requirements for consumer equipment (not allowed to generate interference, have to tolerate whatever there is). Against someone who's expecting to face an environment with a lot of EW, that spray will mostly present targets (an awful lot of the NATO aircraft that have been cruising the Ukrainian border and the Black Sea are basically RF siphons).
More subtle EW, like spoofing signals to take over drones will take a lot more computing, because even if video signals from the drones are't encrypted, the command signals and basic telemetry (position, orientation, altitude, speed) are low bandwidth and can be encrypted without any real operational penalty. We'll probably start to see both encryption and spoofing of that, but it will be very lopsided as to who can spoof and who can't. Local degradation of GPS isn't too hard to do, and we've seen some of that already.
So that leaves the video feed. A key thing to remember about encryption is that it only has to last long enough to not be useful to the eavesdropper when it's cracked. For a drone that might be an hour or two. And if you have a bunch of drones in the air at the same time, that's a bunch of processing power that has to get committed to breaking into the encryption in parallel, and they only have an hour, so the eavedropper has to be in position to get the signal, crack the key, and then get the information to someone for whom it's useful in that time. And the signal for the eavesdropper just has to be degraded enough that they can't tell what it's looking at. The old cable TV "encryption" from the 80s won't quite do it, but it doesn't have to be a whole lot better than that, either. Sure, the eavedropper can record the signal and decrypt it later and might gain some strategic value eventually, but at the tactical level you just need an hour or so of security on the video feed.
As far as lasers and microwaves to defend against autonomous drones- lasers will certainly do it, and there's been plenty of discussion of their limitations (energy hogs, though you can run a pulsed laser that will probably damage an optical sensor off a 9V battery, the tricky part is hitting the camera straight on). Microwaves can also fry the electronics, but need to be on a mobile platform, because the transmitter is going to be a beacon, too.
-
29 minutes ago, Battlefront.com said:
Here's another scenario... EW is up to 11 and no UAS are in the area at all. Ukraine has no idea the column is on the march, so neither FPVs or artillery respond to it. Artillery has no more advantage than UAS in this scenario.
Until the phone rings in Kyiv and it shows a Colorado Springs area code. Kyiv listens for a second, looks at a map, and patches the call through to the BN HQ closest to the column and sets arty and UAS's in motion.
-
8 hours ago, The_Capt said:
That is what made this one weird, it looks more like high mass and high density. Those are very modest frontages and depth the RA is playing with, if it was 50k troops then for the first time in a long time high density mass may have been used effectively. But of course it was at walking speed so the other important factor of momentum was lost. News reports are all indicating that the UA has bottled this one back up, but it is still not clear how (or why) RA dismounted mass was being employed.
I've seen various comments about "50,000", but have we seen any real evidence that it's that many, or just "units which, if they will filled to their full ToE would be 50K, but are only 10% filled" If it were really 50,000 at high density you'd think they'd want to be really,really sure that Ukraine didn't have any tungsten rain available in the area.
-
2 hours ago, The_Capt said:
Mass. The UA had been using small dispersed forces in its tactical bitings last summer. The RA has been using small Cbt teams as short sharp stabs. (and paying for it). From what we can see this was a large dismounted wave with some mech elements. Numbers of 50k have been thrown around. The aim was not infiltration-to-erode, it was to take and hold ground with what looks like unit sized dismounted troops. That is a different kettle of dead Russians.
But isn't that where this has been headed for the past year? Nothing big can move without getting made and then unmade. Small groups of infantry have been effective, or at least hard to target and hit. There are at least two elements to mass: mass and density. The low density has been demonstrated to at least be somewhat effective. Is this attack with mass also at low density, or is it at a high enough density that's going to be relatively straightforward to hit as each element is spotted and turn into a walking meat grinder?
-
1 hour ago, holoween said:
Ok. Good thing space based ISR is so cheap that we can ignore its cost right here. So just to be sure lets check what it costs. 750 million for 5 satelites with the 3 replacements clocking in at 800 million. So for a constallation that hass any chance to track ground movements were already looking at a full brigade equivalent in cost. And its not exactly impossible to shoot them down.
You can do it for a lot lower cost. That's a small constellation of SAR satellites designed for long life (like a decade or more) so as @The_Capt pointed out, it's not really fair to compare it to cost of a tactical ground unit. Especially since you can use it all over the earth for all that time.
PlanetLabs has a couple hundred imaging satellites in orbit that do multiple passes over every place on earth every day at 3.7m resolution. That's not enough to pick out objects easily, but there are ways to squeeze out *much* better resolution in post processing. Those satellites probably cost less than $500K each to build. 20ish of their satellites are ~half-meter optical telescopes that give half-meter resolution on the ground and can do 5-7 visits/day of anyplace on earth. They probably cost less than $100M each to build.
And that's just Planet.
Aside from a few thousand Starlink satellites, there are ~1500 other US based satellites orbiting the earth, including several commercial imaging and SAR companies, plus the NRO/USAF stuff. Add it all up and look at the available resolutions and revisit rates and it's hard to hide anything bigger than a bicycle for very long.
Starlinks cost a few hundred $K each, and they'll sell you a bus (or a truckload of them) that you can slap whatever you want onto (like a telescope and camera for a few hundred more $K) and you can have your own imaging constellation for much, much less than $1B.
Shooting down one satellite is easy. No harder than launching a satellite, really, if you're doing a kinetic kill. Shooting down 1500 satellites is a lot harder.
-
4 hours ago, Anthony P. said:
All of them? And when carrying heavy loads like an RPG-7 grenade?
If you're that certain, I'll bow to your superior wisdom on that specific matter.
Enough of them.
And that's this month's drones. In a few months there will be a whole new generation of them.
Spend a couple dollars and a few hundred grams of lift and you can put both the camera and the warhead on gimbals so you can repoint them both straight down on the fly and land an RPG-7 warhead on whatever part of the tank's top surface you want, no flippy-flight required.
We've also seen no shortage of drones that hover at altitude and drop a shaped charge straight down on their victims. And carrying multiple charges so they can do it more than once if they miss or have enough targets.
-
11 minutes ago, Anthony P. said:
I think you are entirely overestimating the "noise levels" of a few small millimeter wavelength radars, especially considering* the fact that they'll be emitting at <10ft AGL and likely behind more than one physical terrain or man made object (no points for tracking radar emissions from something you've got a direct line of sight to, that would be tech overkill on par with using an X-ray machine to determine if you left your keys in your left pocket or the right one).
*Even more especially when you consider that AESA radars are used in many of the latest APS'. You'd have a fun time eliminating background noise when trying to pinpoint a radar which changes frequeny with every single sweep.
If they're defending against top-attack weapons they have to be transmitting into clear sky.
-
2 hours ago, poesel said:
If radar detection & triangulation is that easy, I'm wondering why anyone in this conflict still has a radar dish?
Because nobody's actually turning them on lest they attract attention?
Detecting a radar transmitter is almost by definition easier than detecting something using the radar. It's how EM fields work.
-
46 minutes ago, poesel said:
No need. These chips are produced in the millions. Only problem is that much of that happens on Taiwan. And the DoD seems to be committed to that already.
I guess you think about that kilowatt class, spinning things? But radar can also be very small. Take this for example (which is not suitable, because the range is only 12m - just for example):
http://hk-suntec.cn/en/product-8871-37864.htmlOutput power is 0.04W. If you can detect that, you can probably already see it with the naked eye.
Ok, I'm just saying, that such a system would be technically feasible and solve the el-cheapo quadcopter problem. But it doesn't exist, and so I'm either wrong or they are still fixing the bugs.
Even with tiny patch antennas you're going to be a beacon for a much longer range than you have a functional radar.
Radar depends on reflectivity, so the radar has a detector that's sensing something like (signal)*(reflectance)*(1/4r^2)ish, where reflectance is some number less than 1. Probably significantly less than 1. I don't even need a better detector than the radar - if I'm using the exact same detector, I get the signal without the reflectivity term and without the 4.
Realistically, if you're generating any kind of periodic signal I can detect it and triangulate on it at a *much* larger range than you can usefully use it as a radar. All I need is a lock-in amplifier and a couple good clocks and I can work down at the background noise limit.
-
5 hours ago, poesel said:
I guess the Ukis would be very, very happy if they could put a 'Shotgun-APS' on every SPG & HIMARS they have, even if that only worked 50% of the time.
Btw: you missed the 'low power radar' part. No point in reaching out for kms if your weapon range is two digit meters.
A low power radar will always look like a beacon for a much longer range than it functions effectively as a radar. An enemy that knows you have them will just toss some sensor/munition package into the mix to fly into your radar until it's gone.
-
4 hours ago, hcrof said:
So I think we will have to disagree on that one. Watching videos of stuff getting blown up in Ukraine makes one think that everything is being watched all at once, and to an extent it is, but how much processing power is required to simultaneously watch thousands of km2 is a level of detail to pick up something within a few minutes?
Not all that much. There are even array detectors that only record changes, not the actual images.
Everything is the same size once you put it on a detetector array (meaning I can use the same tools from microscopy to astronomy because it's all ending up on similar sized arrays and the only difference is scale factor) and change detection is pretty easy. Computers are cheap and happy to sit and watch images all day and all night and alert you with a summary of changes and how likely they are to be interesting to you, and they don't get distracted by something happening in one corner of the image and ignore the rest, or forget to pay attention to the corners.
Having the computer figure out what's actually interesting is trickier, but for a lot of things not all that hard. If you have some idea of the reflection or emission spectra of the things you're looking for and there aren't a lot of other things that match them, it's easy. Even if there are other things that match them there are probably secondary characteristics you can use for filtering. Even cheap, lower power computers can do this. Someone with a bunch of GPUs that burn money faster than they earn [crypto]coin can rent those out for image processing of enormous amounts of ISR data. Or a company that has enough computing power to sell everybody the latest and greatest videogame platform on Black Friday/Cyber Monday without going down can rent that to governments to process all the ISR data there are and still have cycles left over except the day after thanksgiving (that would be the day to stage for your surprise attacks).
Computing is cheap and getting cheaper. ISR is more interesting because with certain capabilities (semiconductor design/mass fab) it becomes very cheap, but without them it's expensive.
-
24 minutes ago, ASL Veteran said:
Everyone said the tank was dead after the ATGM was first brought onto the battlefield. Both sides in Ukraine still seem to want tanks in their inventory so they must think they are valuable for something. To me, the main thing is that if you want your offensive to move at a pace that's quicker than the speed a man can walk, then you will need to have something to transport your infantry in. If your infantry is riding in something, then there will be some incentive to have something that's more powerful than an IFV that you can either use against IFVs or to support an infantry assault.
We all know the old saying about your military preparing to fight the last war only to find out that it doesn't apply anymore, but at the same time I'm not certain how much of what's going on in Ukraine would be applicable to a US force in a peer to peer conflict. Iran and Iraq fought for years and that was basically trench warfare - when the US fought Iraq the first time it looked nothing like Iran Iraq even though many military analysts thought it would. How effective would drones be if the battlefield is fluid with mechanized forces advancing rapidly with an Airforce that has air supremacy? I have to assume that the best military minds are working through the drone problem - if we assume that some sort of a technological or tactical counter can be created, then it seems like the tank is right back in business if you don't want your armies to advance at a walking pace.
I would assume that the Russian and Ukrainian armies operated in a similar way - at least when this whole thing started. Perhaps that's a contributing factor in how things are playing out?
It's not really the drones that are killing the tank, it's the ISR. Drones are (sometimes) the immediate part that blows it up, but the root cause is the ISR.
There are also ATGM crews on quads and motorcycles, artillery with PGMS, artillery that can precision-ish place AT Mines, standoff drones that can paint tanks for laser guided artillery,"smart" uncrewed roadside ATGMs, and eventually @The_Capt's roombas with AT mines, and gawd knows what else. Probably palletized missiles full of air launched smart-ish EFPs that you can push out of a transport plane 50 miles back and paraglide across the FEBA to catch them while they maneuver into place. And their supply trucks, too.
But the key is that you can't move anything the size of a bicycle without commercial systems seeing it within an hour or so. And some states can see if you moved a big-gulp cup, though without quite the same revisit rate. Tanks are big and hot. Even small tanks are big and hot. Electric tanks are big and cold until you move them, but hiding even them from someone with spectral imaging capability isn't easy. And the logistics tail for tanks will also be big and long and hot - you can't move things without leaving a heat trail. So even if you make the tanks like Wonder Woman's invisible plane, the supply chain will lead you right to them. Or the supply chain will get hit so that the tanks can't move.
-
4 hours ago, poesel said:
If drone units get to be 'proper' military units, then a drone platoon is no more than two trucks: one with a 10" container with an openable roof which is filled with drones and their docking bays. This one has to drive into range.
And another truck with communication, command and drone operators. This one just needs to be in com range (by relay) to the drones. They don't need to move that much.
Tracked or wheeled, they can deploy as fast as any mechanized unit.A few billion $ will make a proper military grade equipment out of the hodgepodge setup we have now. Reaction time will be on par with artillery as you can start flying as soon as you know roughly where the target is. Artillery can't be stationary anymore, too.
Careful with those unit abbreviations...
-
2 hours ago, panzermartin said:
Yeah, the numbers and the data seem impossible at the moment. Bear in mind at the same time the size and the many corridors make this city hard to defend also for Ukraine (that has numerical disadvantage) .
I don't think a complete encirclement will be ever possible. But semi encirclement, control of critical junctions, the northern buffer zone and constant harrasment of supply routes are possible and will make the situation difficult for the defender.
How many encirclements has Russia succeeded at so far in the war? They've struggled mightily to fail at encircling even a few km^2 of villages that have a single road intersection. As pointed out by Steve and @The_Capt, Kharkiv is a huge area of very hostile environment to even attempt. And that they failed miserably at when they had at least the pretense of surprise and fresh troops and their best equipment.
1 hour ago, kimbosbread said:If I were the Russians, I’d do my best to make Kharkiv not worth holding for Ukraine. Completely level it a la Grozny. I don’t know if they have enough KABs or artillery within range to do it, but I guess if they were able to hit power and water and take it offline reliably, that would go some towards achieving the effect.
Russia might have an artillery tube advantage, but Ukraine likely has sufficient "action at a distance" and untouchable ISR to limit that. Russia can't just park a few battalions of tubes and send shells for days - they'll be spotted on their way to park and drones will get them within a few rounds if they don't move. So they can probably launch a long steady harassing fire on Kharkiv, but attempting to level it at this point would require tactical nukes, and if Putin hasn't gone there yet, I don't think he's going to go there just to level Kharkiv. If he were considering doing that, why bother with a ground attack?
-
On 4/28/2024 at 1:51 PM, Haiduk said:
Reportedly Ukrianian forces became to use almost noiseless night FPV drones. Here is UKR drone slowly follows for two Russians and despite the range is no more 15-20 m and this is a night, Russians don't hear it.
Drones the size of the smaller FPV drones are remarkably quiet and hard to spot.
I was cycling up in the mountains this weekend and we suddenly heard the whirrrrrr of a drone as it zipped by about 50 feet overhead. We didn't hear it come in, and once it was across the road maybe another 100 feet further it was inaudible. There was a stiff crosswind, but that was literally the only source of noise to mask the drone - the only traffic to speak of was the pickup truck that showed up shortly afterward that shot ahead of us and parked to land the drone on the hood. We didn't see the drone again until it was landing on the hood, even though it was broad daylight and we knew where the person controlling it was sitting.
-
7 hours ago, dan/california said:
Three things about this system, it indicates defense contractors are really working on quadcopter based mine detection, which is a very good thing. The system shown is also a really nifty automated mortar. You are still looking at a LOT of ordinance to clear a lane in minefields the size and density the Russians are laying in Ukraine.
My first take is that it might make more sense to have a drone deposit the approximate equivalent of a DPICM submunition, rather than using the mortar, but this would need detailed analysis of the rate at which each system can take out mines, and how vulnerable it is to being killed while doing so.
Yeah, basically stick a little primer shell onto each mine with a cheap timer or some kind of short range remote control trigger.
-
1 hour ago, kimbosbread said:
I dunno, what about this?
Send a drone withe a room service tray cover to drop on it?
-
6 minutes ago, panzermartin said:
I suspect the main problem for Russians are the myriad simple FPV drones that Ukraine is using. These rely on plain cameras mainly. Smoke can make them less effective at some percent. Guess that would be small. They seem to be producing the smoke on ground level.
I think it could be more effective perhaps with real time satellite intelligence the UA is getting from the West. But I have no accurate idea how "real time" it is in a moving battlefield.
There's not really "real time" continuous satellite coverage. When satellites pass over a region, the data can be sent down in near real time. Orbits of optical satellites (and essentially all satellites) are generally known, though fields-of-regard aren't necessarily, and people worried about space-based ISR have long kept track of when they'd be under a ground track or not. Smoke during an optical pass will provide obscuration, but will be transparent to SAR.
How Hot is Ukraine Gonna Get?
in Combat Mission Black Sea
Posted · Edited by chrisl
I don't know that you even need it to build much of a 3D model - it just needs to do a simple dimensional evaluation: is it flat or not flat? And as Steve points out, you're not limited to visible spectrum single image data. You can ask "does it move/not move?", "is it hot/not hot?", and more without a lot of sophisticated model buildup, and then you feed it to a classifier that decides what to do with it.