Neural networking/AI for B-21

a reply to: thebozeian

A FCS that constantly rewrites itself would be extremely difficult, if not impossible to hack

One would presumably hack an aircraft at the control input lvl. "I want the airplane to go thataway". The flight control laws just decide upon a combination of the hundreds of tiny control -surface movements each second (or other control authority devices) best keep the airplane going "thataway".

a reply to: penroc3

The FCS is already proven to some degree of confidence before flight. It is frequently modified afterwards in response to data which can only be obtained in flight, and generally a gradual expansion of envelope occurs as confidence builds. We are not in a place where simulation and modeling is 100% faithful to the real world.
This sort of neural networking would enable each FCS to be rewriting it's OFP constantly in flight. Which, perhaps I am a curmudgeon, but seems to add a degree of complication and risk to a program they should be pushing with as little of both. The payoff is a marginally better envelope or marginally better control authority at a faster pace -- in theory. I'm skeptical of the risk-reward on this.

There are areas of autonomous operations which will demand this sort of learning on the job AI. I don't think it is necessary in this application which should/will be largely fleshed out and proven already

originally posted by: penroc3
There has been allot of news as far as neural networks go, DARPA even did a TED talk showing how they believe they have simulated a rats brain running on just few server racks.

Neural networks learn by making mistakes, so i would assume they are just training the AI on the ground with supercomputers running millions if not more of flights.

the faster the computer the more flights can be simulated.


imagine a human with millions upon millions of flight hours under every condition conceivable. If the aircraft could maintain a high speed link to a supercomputer on the ground or in space running millions of simulations of the mission they are on it would make a great 'pilot' and would drop the cost per airframe because the supercomputer is on the ground or space somewhere.

I would imagine they separate the data plane from the management plane and keep that system centralized on the ground feeding instructions to the aircraft which are execute via PLC

it's SDN for airplanes

I mean seriously hacking this?

if they encrypt the data on a secure connection how are you going to "hack" this is the software engineers have and network security guys implement proper security controls in the software and over the network?

nobody is going to hack it lol

a reply to: toysforadults

you should ask the ground crews of the F35 and hacking issues


sometimes its the little things that leave back doors open, something overlooked.

nothing is hack proof, its just a matter of how far you are willing to go and the amount of knowledge you have.

a reply to: penroc3

How are you going to exploit aircraft?

What sre you going to do?

a reply to: toysforadults

Theorycrafting here.. but It's entirely possible for a drone dedicated to this as part of electronic warfare. If it can enter the network somehow (?), then it can have access to an aircraft's systems unless they are appropriately firewalled. That is if they already have access to the source code and commands for an aircraft to make it act against its own self interests. We're entering a whole new domain of warfare now and everyone is going to think of anything and everything to gain an advantage.

a reply to: toysforadults

trust me when i say there was a chink in the F35's armor but they fixed it and you had to have access to an air base but the TPTB rushed a fix