Josh Posted January 29 Share Posted January 29 5 hours ago, crazyinsane105 said: Found this very interesting. French engineer somehow gains a Javelin missile and takes it apart. Tech inside is nowhere near high tech and can be easily sourced. Meaning…Russian weapons that rely on semiconductors can’t really be ‘sanctioned’ either. Also means that, chips in ordinary appliances are actually more up to date than what is found in some milspec equipment This has been the case for many decades now. Military hardware gets frozen and repeated for years or decades, and more over does not require a particular high level of processing or memory for most applications. Commercial tech is out of date practically within a year. Most missiles are far, far dumber than an even a several year old iphone. Link to comment Share on other sites More sharing options...
Josh Posted January 29 Share Posted January 29 4 hours ago, KV7 said: The difficulty that can exist is that some specific parts might become unavailable which forces the use of a substitute which takes time to locate or produce. The other problem is if sourcing new parts forces a software change. Link to comment Share on other sites More sharing options...
Samsa Posted January 29 Share Posted January 29 5 hours ago, crazyinsane105 said: Found this very interesting. French engineer somehow gains a Javelin missile and takes it apart. Tech inside is nowhere near high tech and can be easily sourced. Meaning…Russian weapons that rely on semiconductors can’t really be ‘sanctioned’ either. Also means that, chips in ordinary appliances are actually more up to date than what is found in some milspec equipment The pinnacle of this is probably space technology. My Professor for that at university showed us some pretty astonishing stuff in terms of comparing current hardware compared to what is being put in satellites. Though that is of course down to the robustness and reliability requirements in that domain. Mind you, the old chips were still very expensive as they had to go through and ungodly amount of certifications and sometimes hardening iterations. Also possibly Elon changed this up a bit by now... I don't think the chips in things like an ATGM are going to be a problem for Russia, those, as well as (even autonomous) drones just don't do things that require top notch hardware. Maybe they feel the lack of GPUs once they put actual AI in there. I think the lack of chips might, if at all, hit them in some more specialized applications. Or of their production capacity for simple hardware isn't enough for the demand and they can't get enough from China. That is a problem many countires are working on right now after all. Link to comment Share on other sites More sharing options...
Ssnake Posted January 29 Share Posted January 29 The cost for the manufacturer lies in certification, not design and production. The latter will be covered by the buyer. So you freeze, and then never touch anything unless absolutely necessary. Link to comment Share on other sites More sharing options...
KV7 Posted January 29 Share Posted January 29 1 hour ago, Samsa said: I think the lack of chips might, if at all, hit them in some more specialized applications. Or of their production capacity for simple hardware isn't enough for the demand and they can't get enough from China. That is a problem many countires are working on right now after all. This "lack of chips" largely resolves to not everyone getting what they want at the old prices. As most military production is low volume and quite expensive, it can outbid much of the civilian sector. It's not like a washing machine or similar where having to pay $100 more for a circuit board might make it uneconomic to produce. Link to comment Share on other sites More sharing options...
KV7 Posted January 29 Share Posted January 29 1 hour ago, Josh said: The other problem is if sourcing new parts forces a software change. Yes that too. As above it will impose a time delay and use up some engineering capacity but won't be insurmountable. Link to comment Share on other sites More sharing options...
Samsa Posted January 29 Share Posted January 29 17 minutes ago, KV7 said: This "lack of chips" largely resolves to not everyone getting what they want at the old prices. As most military production is low volume and quite expensive, it can outbid much of the civilian sector. It's not like a washing machine or similar where having to pay $100 more for a circuit board might make it uneconomic to produce. Frankly, I have no Insight into what the russians are producing themselves and what they need. Though, I've been part of a few projects where we were delayed for months by not getting the boards we needed and throwing money at the issue didn't solve it. In that sense old hardware is sometimes problematic as there are not always immediate production capabilities available if your standard vendor isn't available anymore due to e.g. sanctions. 15 minutes ago, KV7 said: Yes that too. As above it will impose a time delay and use up some engineering capacity but won't be insurmountable. True, but depending on how much they tailored and optimized their functionalities to the hardware they might as well make it a complete new version of whatever system, which is at least inconvenient I suppose. Link to comment Share on other sites More sharing options...
ink Posted January 29 Share Posted January 29 While bored at work a few days ago, I googled around a bit and found that Chinese academics are publishing papers on autonomous drones for military use. If that kind of thing is making it into the public domain, one has to wonder how much they are working on that in the classified sphere. Same must also be true of the Russians, who are also clearly making some (albeit limited) progress with Lancet. Link to comment Share on other sites More sharing options...
seahawk Posted January 29 Share Posted January 29 Everybody is working on that. Link to comment Share on other sites More sharing options...
ink Posted January 29 Share Posted January 29 9 minutes ago, seahawk said: Everybody is working on that. Yes, fair point, young man! However, it's now becoming increasingly pertinent, isn't it? Link to comment Share on other sites More sharing options...
Ssnake Posted January 29 Share Posted January 29 57 minutes ago, ink said: While bored at work a few days ago, I googled around a bit and found that Chinese academics are publishing papers on autonomous drones for military use. If that kind of thing is making it into the public domain, one has to wonder how much they are working on that in the classified sphere. Anyone trusting that humanity will not develop autonomous killbots (because people will recognize how problematic this is) is, frankly, delusional. They might not come this war or the next, but they will come before the decade is out. They might not be very good in the beginning either, but each generation will improve. Link to comment Share on other sites More sharing options...
seahawk Posted January 29 Share Posted January 29 38 minutes ago, ink said: Yes, fair point, young man! However, it's now becoming increasingly pertinent, isn't it? Yes, especially as it has become easy, as people do not know how old fashioned classic military hardware is. A recent middle class smartphone has more processing power than most military hardware. And we are just seeing the KI specified chips growing into the mainstream market. The software is also basically finished, at least if you are willing to work with about 10% error. Link to comment Share on other sites More sharing options...
Roman Alymov Posted January 29 Share Posted January 29 Reportedly new Geran' warhead, with shaped charge and incendiary elements https://t.me/infomil_live/3574 Link to comment Share on other sites More sharing options...
bojan Posted January 29 Author Share Posted January 29 (edited) 3 hours ago, Samsa said: ...I don't think the chips in things like an ATGM... As long as ATGM is SACLOS vs F&F you don't need a single chip for it to work. Early SACLOS ATGMs (TOW, Fagot/AT-4, SS-11 in SACLOS version etc) did it with more-less radio shack level of electronics. Only consequence was somewhat bulky guidance unit. Edited January 29 by bojan Link to comment Share on other sites More sharing options...
ink Posted January 29 Share Posted January 29 32 minutes ago, Ssnake said: Anyone trusting that humanity will not develop autonomous killbots (because people will recognize how problematic this is) is, frankly, delusional. They might not come this war or the next, but they will come before the decade is out. They might not be very good in the beginning either, but each generation will improve. My feeling, from talking to people familiar with AI tech and from reading a bit of that Chinese public-domain research, is that with an expensive sensor (mmw or TI), autonomous missions are already possible against vehicle-sized targets. However, doing that just based on visual-range camera images is a good way off (apparently). Autonomous drones will - when deployed in numbers - be a real bitch. There'll be no signal to jam, they'll probably be cheap as chips, and they will be able to operate at longer ranges. It gets even scarier when you start to think about them going after personnel-sized targets... Potentially in swarms. Link to comment Share on other sites More sharing options...
seahawk Posted January 29 Share Posted January 29 Totally not a long way off. You just need to look at all the modern assistance systems in cars. What is a front assist for you car can easily become the guidance section for a drone. The challenges still is automatic target detection with a meaningful range, as the FoV for optical sensor is limited if the range is sufficient. So you would combine it with a MMW or something. But that is for loitering munitions. If you just want a smart missile that flies to a given set of co-ordinates and attacks a target there, all you need is a smartphone. Link to comment Share on other sites More sharing options...
ink Posted January 29 Share Posted January 29 (edited) 6 minutes ago, seahawk said: Totally not a long way off. You just need to look at all the modern assistance systems in cars. What is a front assist for you car can easily become the guidance section for a drone. The challenges still is automatic target detection with a meaningful range, as the FoV for optical sensor is limited if the range is sufficient. So you would combine it with a MMW or something. But that is for loitering munitions. If you just want a smart missile that flies to a given set of co-ordinates and attacks a target there, all you need is a smartphone. As I understand it, and as I've already said on here multiple times, target recognition is still too difficult for autonomous detection and selection. Self-driving cars struggle to operate in a highly rules-heavy environment (i.e. roads)* where the 'targets' go to great lengths to signal their presence. Doing something similar in an environment where the target is actively trying to hide** is still not something that can be done reliably (using just visible light cameras). * See Tesla's recall in China. ** Not to mention natural cover, smoke, weather... Edited January 29 by ink Link to comment Share on other sites More sharing options...
seahawk Posted January 29 Share Posted January 29 No, you need sensor fusion, but that is exactly what cars do. https://www.ti.com/sensors/mmwave-radar/automotive/overview.html Link to comment Share on other sites More sharing options...
Samsa Posted January 29 Share Posted January 29 23 minutes ago, ink said: As I understand it, and as I've already said on here multiple times, target recognition is still too difficult for autonomous detection and selection. Self-driving cars struggle to operate in a highly rules-heavy environment (i.e. roads)* where the 'targets' go to great lengths to signal their presence. Doing something similar in an environment where the target is actively trying to hide** is still not something that can be done reliably (using just visible light cameras). * See Tesla's recall in China. ** Not to mention natural cover, smoke, weather... Yes and no. Some things are easier on roads, but generally, navigating a ground environment with lots of other actors in it is harder than flying. Also obscured or partly obscured vehicles are a big deal and something we are still actively working at. The real trouble are pedestrians, animals and what ever else (often called "Vulnerable road users). Those are often obscured, move erratically and produce tons of non standard situations. Detecting those reliability is still a huge topic. Contrast that to the other extreme where any army might go "Well its warm and its on their side of the Frontline, just blow it up". If you want proper ID, that's different story, but I wouldn't say more difficult than what a Tesla might have to deal with. Link to comment Share on other sites More sharing options...
Roman Alymov Posted January 29 Share Posted January 29 27 minutes ago, ink said: As I understand it, and as I've already said on here multiple times, target recognition is still too difficult for autonomous detection and selection. Self-driving cars struggle to operate in a highly rules-heavy environment (i.e. roads)* where the 'targets' go to great lengths to signal their presence. Doing something similar in an environment where the target is actively trying to hide** is still not something that can be done reliably (using just visible light cameras). "Too difficult" is not mumerical metrics. We could better talk about % of errors. Now the production cost of FPV drone in Russia* is about USD500 - it is, as far as i understand, about the cost of Russian-made 152mm shell, or ~1/10 of a cost of Western-made 155mm shell. So even is we assume only 10% of AI-guided FPVs will find real human targets in enemy trenches - it will mean that typical forest belt position defended by typical 10 men would be wiped out by only 100 FPVs launched - in monetary terms, equivalent of 10 Western shells fired or 100 Russian 152mm -and note guns wearout etc. not included into calculation. * Yes, this production is, de-facto, assembly of Made in China components on Made in Russia frame - the same as in Ukraine and West. Link to comment Share on other sites More sharing options...
bojan Posted January 29 Author Share Posted January 29 (edited) 5 minutes ago, Samsa said: ...Contrast that to the other extreme where any army might go "Well its warm and its on their side of the Frontline, just blow it up"... Problem with "go for biggest hotspot" (that was tried for various ammunitions) is that all drones in the sector will probably go for same, as after first hit it will be biggest hotspot. So first drone has to go for biggest hotspot, second for 2nd biggest, 3rd for 3rd etc. Which needs at least some logic programmed into target selection. Edited January 29 by bojan Link to comment Share on other sites More sharing options...
Roman Alymov Posted January 29 Share Posted January 29 2 minutes ago, bojan said: So first drone has to go for biggest hotspot, second for 2nd biggest, 3rd for 3rd etc. Which needs at least some logic programmed into target selection. There is a common belief that this task of target selection/target distribution was allready solved on old Soviet Kh-22/Kh-32 missiles, with their ancient electronics. Link to comment Share on other sites More sharing options...
seahawk Posted January 29 Share Posted January 29 2 minutes ago, bojan said: Problem "go for biggest hotspot" is that all drones in the sector will probably go for same, as after first hit it will be biggest hotspot. So first drone has to go for biggest hotspot, second for 2nd biggest, 3rd for 3rd etc. Which needs at least some logic programmed into target selection. Inter-swarm communication and sensor sharing would be very advanced. But at the moment you can works with separation either by distance or time. Link to comment Share on other sites More sharing options...
Ssnake Posted January 29 Share Posted January 29 First scout and register approximate locations with a small swarm, then attack spotted targets with a bigger swarm. Exclude point targets with measured heat exceeding 200°C. Link to comment Share on other sites More sharing options...
ink Posted January 29 Share Posted January 29 42 minutes ago, seahawk said: No, you need sensor fusion, but that is exactly what cars do. https://www.ti.com/sensors/mmwave-radar/automotive/overview.html But that's pricey, given you're blowing up all your sensors every time. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now