Indian Navy Developments & Discussions

Bleh · Nov 21, 2024

Blood+ said:
True but then those same external factors will apply to the human pilot as well and dare I say, the human will be feel their effect to a greater degree.

Nope... In a simulation human doesn't exactly know the other jet's location is. In say, DCS you'll have to look for the key just like real life.
But if your playing against a bot that's part of the simulator, it'll have access to all the data about your position direction speed. Odds will even out more in real scenario. If the AAM's programming is any benchmark, AI has a long way to go.

AI's advantage is processing power. That makes it very good at chess because it can predict all permutations & combination of your possible moves allotted for each piece.
Now if you bring more chaos it, imagine if all pawns were queens. That'd make it harder for the system to dominate simply because the outcomes to process are much more.

In real like combat scenario the option simply are to high. AI itself will fly better than a human pilot, perfect turns at perfect speeds. But a human pilot will deviate. So unless the guy is flying absolutely perfectly like the bot, it'll have to be reactive not proactive. That's how to bring it to do stuff you want it to do. Like how notching is done, you do something to make the AAM react in a specific way & make use of that to misguide it.

randombully · Nov 21, 2024

Best way would be AI assisting the human by providing info and predicting, like in a dog fight AI can track the jet much better than the eye or human pilot it can show the position to the pilot then the decision remains in pilots hands on what action to take.
AI lacks the creativity of human minds so unpredictability will left to humans and final decision of action too.

Bleh said:
Nope... In a simulation human doesn't exactly know the other jet's location is. In say, DCS you'll have to look for the key just like real life.
But if your playing against a bot that's part of the simulator, it'll have access to all the data about your position direction speed. Odds will even out more in real scenario. If the AAM's programming is any benchmark, AI has a long way to go.

AI's advantage is processing power. That makes it very good at chess because it can predict all permutations & combination of your possible moves allotted for each piece.
Now if you bring more chaos it, imagine if all pawns were queens. That'd make it harder for the system to dominate simply because the outcomes to process are much more.

In real like combat scenario the option simply are to high. AI itself will fly better than a human pilot, perfect turns at perfect speeds. But a human pilot will deviate. So unless the guy is flying absolutely perfectly like the bot, it'll have to be reactive not proactive. That's how to bring it to do stuff you want it to do. Like how notching is done, you do something to make the AAM react in a specific way & make use of that to misguide it.best

Ayan Barat · Nov 21, 2024

Bleh said:
That'd make it harder for the system to dominate simply because the outcomes to process are much more.

randombully said:
AI lacks the creativity of human minds so unpredictability will left to humans and final decision of action too.

At this point I can just feel sad

A simple prompt of "You're piloting a fighter and a human is piloting the same. You both come inside the WVR of each other. Ways in which you can defeat the human pilot in a simulated dogfight" and here's AI for you

And to my utter surprise, he topped it off with an actual step by step strategy.

Now few points to consider
> This is vanilla ChatGPT and that too the free one. In an actual fighter you'd have a purpose built, way more refined AI model.
> Hardware specs here was Snapdragon 712/4GB RAM with most of work happening over net. In actual combat you'd be having top of the line Nvidia GPUs doing everything.
> There was absolutely no training data and still this f*cker managed to come up with shortcomings of human pilot just based on what was available in open source. In real life you'd have TBs worth of confidential data from both simulators and actual mock dogfights.

shade2 · Nov 21, 2024

Royal Navy finished as full-spectrum force as Albion and Bulwark axed

The UK Royal Navy's amphibious assault capability has ended with the scrapping of HMS Albion and Bulwark a decade ahead of schedule.

www.naval-technology.com

These are LPDs build in the 2000s, so they are 20+ years old ships
Folks in the UK thread were saying we should buy these

thoughts of y'all nibbiars?

mist_consecutive · Nov 21, 2024

View: https://x.com/NewsIADN/status/1859563629202289152

Gslv Mk3 · Nov 21, 2024

Bleh · Nov 21, 2024

Ayan Barat said:
At this point I can just feel sad

A simple prompt of "You're piloting a fighter and a human is piloting the same. You both come inside the WVR of each other. Ways in which you can defeat the human pilot in a simulated dogfight" and here's AI for you
View attachment 15992
View attachment 15993
View attachment 15994
View attachment 15995
View attachment 15996
View attachment 15997
And to my utter surprise, he topped it off with an actual step by step strategy.
View attachment 15998
Now few points to consider
> This is vanilla ChatGPT and that too the free one. In an actual fighter you'd have a purpose built, way more refined AI model.
> Hardware specs here was Snapdragon 712/4GB RAM with most of work happening over net. In actual combat you'd be having top of the line Nvidia GPUs doing everything.
> There was absolutely no training data and still this f*cker managed to come up with shortcomings of human pilot just based on what was available in open source. In real life you'd have TBs worth of confidential data from both simulators and actual mock dogfights.

Some underlined text from an article mean little & less.

Jets themselves are rated for 10g... Wings will break off after.
And it's not like AI's jet won't bleed energy. I comment needed in my research only. Even computer chess bots have become more beatable after human players have deducted how they "think".

Maybe there will be platforms that fits to exploit fully the AI's capabilities like a 20g rated jet, but they'll have to be specially developed. Time enough for that. AI also will have to develop much.
Otherwise we'd be having invincible longranged suicide Drone-AAMs that can pull 40g doing CAPs at Mach3, instead of actual jets. We don't because they're beatable using there usual bot brain weaknesses

VictorDelta4 · Nov 22, 2024

Ayan Barat · Nov 22, 2024

Bleh said:
...otherwise we'd be having invincible longranged suicide Drone-AAMs that can pull 40g doing CAPs at Mach3, instead of actual jets. We don't because...

We don't because it's hardly been four years since we've realised the true potential of AI in dogfighting. Until the news of an AI-Agent beating human pilot in AlphaDogFight has broken in mid-2020, nobody would have imagined something like this is even possible in real world.

Think about it, exactly how many air-to-air combat focused drone program started in this period? Almost all of the loyal wingman projects (theoretically the best platform to use a similar system) were already past atleast their prototype phase when this news broke...XQ-58 was flying in 2019, MQ-28 was flying by early 2021. It's (almost) impossible to modify a slow, bomber type UCAV into an extreme high performance dogfighting one so the only option is to start from scratch. And for that you need time, time to start a program, choose two firms, evaluate them...but perhaps the most difficult; sanctioning budget from the Congress FOR A PLANE THAT CAN AUTONOMOUSLY TARGET ENEMY AIRCRAFT USING AI. It's scary for civilian politicians
So give it a bit time and trust me, it'll definitely happen. The next "batch" of loyal wingman would be CAP specialists

The reason jet aircrafts are rated to just 10g is not because of how fragile aluminium is but it's rather because there's no point in making a 10g+ aircraft when the pilot itself would most probably pass out at 9. What's the point of a 10' high door when the tallest family member is 6'!? Wings are not going to break off after 10g. Just search any stunt plane and you'd see the max limit being 12g. In 1960, we're making manned aircrafts capable of going hypersonic. It's just the fact that as of now we really had no reason to push the boundaries of a plane...but after 2020, we now have. And as for the flaws of AI, well here we're discussing this there they're currently going through training data

Blood+ · Nov 22, 2024

Bleh said:
Nope... In a simulation human doesn't exactly know the other jet's location is.

You didn't even watch the video.

Bleh said:
In say, DCS you'll have to look for the key just like real life.
But if your playing against a bot that's part of the simulator, it'll have access to all the data about your position direction speed. Odds will even out more in real scenario.

Yes, precisely!! And it'll play out the same in real life as well!!

Bleh said:
If the AAM's programming is any benchmark, AI has a long way to go.

AAM's do not use AI, not yet anyhow.

Bleh said:
AI's advantage is processing power. That makes it very good at chess because it can predict all permutations & combination of your possible moves allotted for each piece.
Now if you bring more chaos it, imagine if all pawns were queens. That'd make it harder for the system to dominate simply because the outcomes to process are much more.

Yeah, you're not really making a case here, my guy!! Because those same handicaps would apply to a human grandmaster as well, only exacerbated a hundred fold that is.

Bleh said:
In real like combat scenario the option simply are to high. AI itself will fly better than a human pilot, perfect turns at perfect speeds. But a human pilot will deviate. So unless the guy is flying absolutely perfectly like the bot, it'll have to be reactive not proactive. That's how to bring it to do stuff you want it to do. Like how notching is done, you do something to make the AAM react in a specific way & make use of that to misguide it.

And what if they comes up with a fighter design that can pull, let's say 20g or above?? What then??

Blademaster · Nov 22, 2024

Ayan Barat said:
We don't because it's hardly been four years since we've realised the true potential of AI in dogfighting. Until the news of an AI-Agent beating human pilot in AlphaDogFight has broken in mid-2020, nobody would have imagined something like this is even possible in real world.

Think about it, exactly how many air-to-air combat focused drone program started in this period? Almost all of the loyal wingman projects (theoretically the best platform to use a similar system) were already past atleast their prototype phase when this news broke...XQ-58 was flying in 2019, MQ-28 was flying by early 2021. It's (almost) impossible to modify a slow, bomber type UCAV into an extreme high performance dogfighting one so the only option is to start from scratch. And for that you need time, time to start a program, choose two firms, evaluate them...but perhaps the most difficult; sanctioning budget from the Congress FOR A PLANE THAT CAN AUTONOMOUSLY TARGET ENEMY AIRCRAFT USING AI. It's scary for civilian politicians
So give it a bit time and trust me, it'll definitely happen. The next "batch" of loyal wingman would be CAP specialists

The reason jet aircrafts are rated to just 10g is not because of how fragile aluminium is but it's rather because there's no point in making a 10g+ aircraft when the pilot itself would most probably pass out at 9. What's the point of a 10' high door when the tallest family member is 6'!? Wings are not going to break off after 10g. Just search any stunt plane and you'd see the max limit being 12g. In 1960, we're making manned aircrafts capable of going hypersonic. It's just the fact that as of now we really had no reason to push the boundaries of a plane...but after 2020, we now have. And as for the flaws of AI, well here we're discussing this there they're currently going through training data

Not true. An AC-130 Ghostrider combat plane was inverted after a mishap and even though the plane made it in one piece and landed safely, it was written off as a total loss because the inversion outg'ed the g limits of the plane.

Brand New AC-130 Ghostrider A Total Loss After Going Inverted While Out Of Control

One of the Air Force Special Operations Command’s brand-spanking-new AC-130J Ghostrider Gunships has to be scrapped due to a test flight that went horribly awry. Luckily nobody was harmed but the $115 million dollar highly-modified Super Hercules will never fly again.

jalopnik.com

Bleh · Nov 22, 2024

Blood+ said:
You didn't even watch the video.

And what if they comes up with a fighter design that can pull, let's say 20g or above?? What then??

Invincible longranged suicide Drone-AAMs that can pull 40g doing CAPs at Mach3, instead of actual jets.

Oh wait ..

Blood+ · Nov 22, 2024

Bleh said:
Invincible longranged suicide Drone-AAMs that can pull 40g doing CAPs at Mach3, instead of actual jets.

Oh wait ..

It's been JUST THREE FUCKING YEARS!! Did you just start sprinting the moment you came out of your mother's womb?? Give this tech some time to mature!!

Ayan Barat · Nov 22, 2024

Blademaster said:
Not true. An AC-130 Ghostrider combat plane was inverted after a mishap and even though the plane made it in one piece and landed safely, it was written off as a total loss because the inversion outg'ed the g limits of the plane.

Aayein!

There's a difference of day and night between AC-130 and a fighter.
The aspect ratio is perhaps the complete opposite of one another, wing loading is different, the turning moments would be different, the 1t outboard engine assembly on a C-130 would weigh way more than a 1t bomb on a F-15's outboard pylon because of the distance it's from the centreline, the moment of inertia would act more on one's wings than other's when they abruptly stop after doing a roll...

But perhaps the easiest way to state the differences between them would be a side by side comparison of C-130J and Su-30. Also worth noting is the fact that Su-30 (F-15 too) are absolutely humongous platforms compared to other fighters and yet they're such puny in comparison to C-130.

Now leave aside all the science and just answer from a layman's perspective.
What will be easier to break or bend; a long narrow thin "strip" or a short wide thick "slab"?

It all comes down to the fact that for what purpose a plane was designed. Try to do an arrestor hook landing with a light trainer and the whole plane would be left mangled, yet F-18s do it hundreds of times before even needing an inspection.

Ayan Barat · Nov 22, 2024

Bleh said:
Invincible longranged suicide Drone-AAMs that can pull 40g doing CAPs at Mach3, instead of actual jets.

Oh wait ..

If I may quote Taylor Swift's Who's Afraid of Little Old Me?

But what if they did?

None of what you mentioned is even hard, let alone impossible from a technological point of view.
The electromechanical assemblies used in guided artillery shells like Excalibur operates above 10,000g
It's been almost 60 years since X-15 went beyond Mach 6 and that too with a human pilot. Again 60 years since MiG-25s started flying repeatedly at Mach 3

Can USA/Russia make a 100MT fusion bomb? Absolutely yes; they've already made 25-50MT ones back in 1960s
But is there a point in making one when you already have things like ICBMs and MIRV?

It's is same thing with this too, there was no point in developing something similar. But after 2020 there's. I can bet you that in just a couple of years you'd find a patent by Lockheed Martin or a research paper from a Chinese aeronautics university related to what we're discussing today

Bleh · Nov 22, 2024

Ayan Barat said:
If I may quote Taylor Swift's Who's Afraid of Little Old Me?

But what if they did?

None of what you mentioned is even hard, let alone impossible from a technological point of view.
The electromechanical assemblies used in guided artillery shells like Excalibur operates above 10,000g
It's been almost 60 years since X-15 went beyond Mach 6 and that too with a human pilot. Again 60 years since MiG-25s started flying repeatedly at Mach 3

Can USA/Russia make a 100MT fusion bomb? Absolutely yes; they've already made 25-50MT ones back in 1960s
But is there a point in making one when you already have things like ICBMs and MIRV?

It's is same thing with this too, there was no point in developing something similar. But after 2020 there's. I can bet you that in just a couple of years you'd find a patent by Lockheed Martin or a research paper from a Chinese aeronautics university related to what we're discussing today

Kinda my point... If the AI develops in this aspect I think would have this sort of long range smart suicide-AAMs before fighters lobbing dumber SAMs that can be notched & spoofed.

vampyrbladez · Nov 22, 2024

Indian Navy submarine & fishing vessel collide off Goa coast, search for 2 missing fishermen on

Defence sources told ThePrint the vessels crashed into each other around 9.30 pm Thursday evening when the submarine was making a transit between two ports.

Sources said full details of the accident were yet to emerge, but it looks there were no major damages to the vessel as it continued to sail. They said the Naval headquarters were waiting for the submarine to reach its destination to assess damage and the reasons behind the collision.

https://theprint.in/defence/indian-...st-search-for-2-missing-fishermen-on/2368415/

shade2 · Nov 22, 2024

vampyrbladez said:
Indian Navy submarine & fishing vessel collide off Goa coast, search for 2 missing fishermen on

https://theprint.in/defence/indian-...st-search-for-2-missing-fishermen-on/2368415/

What damage can a 2rs fishing boat do to a submarine?

Ayan Barat · Nov 22, 2024

shade2 said:
What damage can a 2rs fishing boat do to a submarine?

Weird question

At the least rip off the anechoic tiles; not expensive tho as they're usually replaced after few deployments. Break all the masts; partial mission kill with huge costs in repair. At most hit and damage the propeller; even mobility kill with MLR level of costs.

shade2 · Nov 22, 2024

Ayan Barat said:
Weird question

At the least rip off the anechoic tiles; not expensive tho as they're usually replaced after few deployments. Break all the masts; partial mission kill with huge costs in repair. At most hit and damage the propeller; even mobility kill with MLR level of costs.

I expected that the masts might break but didn't know about the rest, no idea wooden fishing boat could do this much damage

Hope the sub has sustained minimal damage

Indian Navy Developments & Discussions

Laughing Bulldog

Member

Full Member

Senior Member

Golgappe Expert

Member

Laughing Bulldog

Senior Member

Full Member

Member

Full Member

Laughing Bulldog

Member

Full Member

Full Member

Laughing Bulldog

Member

Senior Member

Full Member

Senior Member

Similar threads

Trending Threads

Share this Site