The X-BAT UCAV from Shield AI

[Signal Witch]

If I’m not mistaken, much of the typical F-117 mission was/is controlled by a prerecorded automated flight management system. A large part of the flight path and mission were essentially pre-recorded and then run onboard for the actual flight. That seems to be a rudimentary approach to what is now being attempted here and in other UAVs, no? Since quite a large portion of those Nighthawk missions could be automated, the pilot wasn’t having to focus on as many small things and could concentrate on other critical demands. This all just sounds like the latest iteration of that same basic idea.

They’ve simply managed to slowly reduce the pilot’s input to the point that a pilot is no longer required.

Avicula,

Yeah, that’s pretty much it. The F-117’s flight control and navigation systems handled most of the grunt work once the mission was programmed. The pilot was basically there to monitor, manage the radar picture, and take over if something went sideways. That was automation, not decision-making.

 What’s changing now is the aircraft’s ability to handle the unexpected. The old systems just executed a plan. X-BAT and similar platforms are being built to recognize when the plan no longer works and then decide what to do next without waiting for input from a pilot or ground station.

 So it’s the same family tree, just a few generations down the line. The Nighthawk could fly itself along a path, but it couldn’t think its way out of trouble. These new systems are learning how to do that, which is why Shield keeps calling it autonomy instead of automation.

[Zaphod58]

Unmanned f-16s as well as F-18s and I would not be surprised if one or more F-117s have flown unmanned. Replying for Sirius... Unmanned autonomous everything is a "THANG" now and has been for several years. .. ! 
https://www.militaryaerospace.com/commer...g-growlers

Zombie Vipers are target aircraft not capable of combat. The unmanned Growler was a demonstration, using software for Loyal Wingman and isn’t going to go anywhere.

[EXETER]

Watch the video where it is stated 20+ and in some cases up to 30 Gs  An off the shelf Extra 300 can do 10 Gs with no airframe problems. Extra 300 is rated for +10 -4 and I have heard back in the day with some modified that could do even more.

As you know it is mostly the limits of the pilot not the machine that sets the G limits. Why build something that can pull 30+ when there is never going to be a living creature on-board.

As an aerospace engineer I should comment here.

It is relatively easy to design a structure for more than 10 Gs if, for some reason, you wanted to.  But a fighter would probably not be designed for more than about that amount of turning ability whether it was manned or not.  

Fighter aircraft typically have an L/D (Lift to Drag ratio) of about 10.  So if you are flying along straight and level, the wings are generating 1 G of lift. That means the jet engine has to provide an amount of thrust equal to only 1/10 the weight of the plane. If the wings are suddenly called on to produce 10 Gs of lift in a maneuvering dogfight, the drag will suddenly go up to an amount approximately equal to the weight of the aircraft. Frontline fighters like the F-15, for example have a thrust to weight ratio of about 1, or slightly higher. So even if the airframe was designed to withstand a higher G loading, the engine thrust could not sustain that turn rate, which means that the aircraft would lose speed very quickly and become a sitting duck.

The G-limit is limited by the thrust to weight ratio of the aircraft.

[Signal Witch]

As an aerospace engineer I should comment here.

It is relatively easy to design a structure for more than 10 Gs if, for some reason, you wanted to.  But a fighter would probably not be designed for more than about that amount of turning ability whether it was manned or not.  

Fighter aircraft typically have an L/D (Lift to Drag ratio) of about 10.  So if you are flying along straight and level, the wings are generating 1 G of lift. That means the jet engine has to provide an amount of thrust equal to only 1/10 the weight of the plane. If the wings are suddenly called on to produce 10 Gs of lift in a maneuvering dogfight, the drag will suddenly go up to an amount approximately equal to the weight of the aircraft. Frontline fighters like the F-15, for example have a thrust to weight ratio of about 1, or slightly higher. So even if the airframe was designed to withstand a higher G loading, the engine thrust could not sustain that turn rate, which means that the aircraft would lose speed very quickly and become a sitting duck.

The G-limit is limited by the thrust to weight ratio of the aircraft.

 
EXETER,

That’s a solid explanation and it matches what I’ve heard from the engineering side. You’re right that thrust-to-weight is what really limits sustained G, not just what the structure can take. A jet can only fight drag for as long as it has the power to hold energy.
 
Also, you may be able to confirm this, but from what I understand, most manned fighters like the F-15 are rated for about +9 G as their load limit. Structurally they are built to survive about 1.5 times that, around +13.5 G, which is called the ultimate load. That is the point the airframe is expected to withstand once without failing, not something you would ever try to sustain. Past that you start bending spars and overstressing the frame fast.
 
Even if you removed the pilot and let the system fly itself, you would only gain a little margin before physics takes over. A QF-15 might spike 10 or 11 G for an instant, but holding anything near 13 G would risk tearing the airframe apart. The limits are still set by thrust, drag, and material strength, not by who is sitting in the cockpit.

[Blackfingers]

 So it’s the same family tree, just a few generations down the line. The Nighthawk could fly itself along a path, but it couldn’t think its way out of trouble. These new systems are learning how to do that, which is why Shield keeps calling it autonomy instead of automation.

Thats why I think a variation of Boolean code (true/false) is being used as a base code.I mentioned Gcode as its a set of "automated" codes (move x,y,z) that is used by CNC or 3D printers.
Remember Grey code in washing machines becoming a thing where sensors would inform the machine if clothes were clean or not.
Hey Exeter Aircraft sheetmetal worker here great to hear from one from "the other side" :)

[Sky727]

1150/2000 mile range, can complete mission without GPS and do away with the pesky need for a runway even with a combat load.  Seems to be limited to 4G maneuvers  

 
Shield AI currently plans to begin VTOL and then further flight testing in the all of 2026 with the aircraft ready for service by 2028 and in production starting in 2029. And that is a very

Typical long lead time for anything defense related. Production to begin in 2029 ....We will just have to wait and see if the Rah Rah matches the actual aircraft performance considering it has not even flown VTOL as of yet.

[Sirius]

[Signal Witch]

Thats why I think a variation of Boolean code (true/false) is being used as a base code.I mentioned Gcode as its a set of "automated" codes (move x,y,z) that is used by CNC or 3D printers.
Remember Grey code in washing machines becoming a thing where sensors would inform the machine if clothes were clean or not.
Hey Exeter Aircraft sheetmetal worker here great to hear from one from "the other side" :)

You’re right. Boolean logic still underpins everything. The difference now is the number of layers built on top of it.

The older systems relied on simple true/false decision chains, while the newer ones use weighted logic combined with sensor fusion and realtime learning loops. The foundation is the same, but th architecture has grown much deeper.

[Signal Witch]

Found these renderings of the X-Bat from Shield, thought it was interesting...

[Blackfingers]

Nice platforms.With folded wings I can see them fit into a standard small shipping container.Next day Delivery by Amazon to a third world theatre of operations. :)

The older systems relied on simple true/false decision chains, while the newer ones use weighted logic combined with sensor fusion and realtime learning loops. The foundation is the same, but th architecture has grown much deeper.

I think biggest trap is getting caught in a logic loop due to a faulty image.
Russia tried to use Ai imaging on some of its Drones last year off memory with limited success.
Non GPS being used?Looks like Qantum navigation is a lot more advanced then whats been reported.