TUCKnTRUCK
That guy
Back when I was getting into ski patrol, had to learn knots. The old adage of “If you can’t tie a knot, tie a lot” feels like it would apply to manual bombs as well.
Hackers are taking over planes’ GPS — experts are lost on how to fix it
- Thread starter Oxman
- Start date
Not taught anymore. Used to be a qual event back in my day in the Hog, a pride in doing them accurately. Pointy nose jets haven’t done it in a long time. Don’t qual in it, you aren’t combat qualed. For a while now, it’s only been a one-time fam event…a “so this is what dropping a manual bomb looks like, something that you will never do” event. More overreliance on GPS and laser systems. Though we have done this to ourselves, really. Funny, I had to drop manual bombs next to a friendly camp under siege after all the computed symbology crapped out on me one night,
arkflyr
Well-Known Member
That was the exact thing I was told when I became a patroller.
TUCKnTRUCK
That guy
I mean, drop enough and I’m sure somebody will hit something…
ppragman
FLIPY FLAPS!
I actually suspect pilotless airliners are a lot "easier" (with huge air quotes) of a problem. Like, it's really cool what they can do, the tech is... bad ass in every conceivable way. But driving is a crazy, dynamic environment. There's a whole group of people who are working on the psychology involved in it all and trying to model that too.So about the same time as pilotless airliners?
Like, imagine you're waiting to walk across the street. You make eye contact with the driver, he sees you, you see him. How close can you reasonably get to the car now? Humans will get really close even if the vehicle is still moving... really close. Once you've made eye contact there's a social contract made, and you know that the driver won't run you over, so you can get extremely close to the road. This screws with self-driving vehicles, because humans are comfortable being closer to a vehicle or even starting out into the street after they've locked eyes with the driver. How does a robot handle that? And that's just 1 problem.
Largely, I think the result will be, "These should be all on rails so people can't get run over and accidents can't happen" and tech will have "re-invented" the train.
The conference was great, but I remember reading several posters and thinking, "oh, I'm the only airplane person here, because this is a problem we solved in aviation like 40 years ago..."
I think we'll get it right eventually (hopefully) - I think legal responsibility for the performance of your self-driving vehicle will be what drives capability not pure tech. I think the "right" path is what Mercedes-Benz is doing presently. Small iterative steps are better than "all at once" in some applications.
What it's like to experience Mercedes-Benz's Drive Pilot automated system
Mercedes-Benz is the first automaker to bring an SAE Level 3 system to the U.S. market.
abcnews.go.com
bike21
9-5 Ruins Lives
At my hill it was less rope work and more "hey you in the Carhartts, PIZZA!!"
Lawman
Well-Known Member
True, but we don’t have enough (nor can we build enough) to do that for very long.
Sent from my iPhone using Tapatalk
arkflyr
Well-Known Member
Being in Montana that was for the UND hockey jerseys and hunter’s camouflage.
“Oh god that Jerry is going to drop in there, please don’t…”
DE727UPS
Well-Known Member
I never needed to be that smart when I went to Riddle, thank goodness.
bike21
9-5 Ruins Lives
We had a single black run that emptied into a cornfield if you happened to not make any turns. It was also known as the Jerry collection point and you'd often see Carhatts with skis slung over the shoulder walking back toward the lift.
Roger Roger
Bottom of the list
I don’t ski but I still find jerry slander to be hilarious.
Doubly so because of jerry w
arkflyr
Well-Known Member
Jerryoftheday is a treasure of Instagram fun.
My old mountain opens the Friday after thanksgiving generally with artificial snow on the lower mountain only, ribbon of death stuff. There is always some numbnuts in full send mode going straight down through it that has to be chased down and corrected.
It happens everywhere though I’m sure.
My old mountain opens the Friday after thanksgiving generally with artificial snow on the lower mountain only, ribbon of death stuff. There is always some numbnuts in full send mode going straight down through it that has to be chased down and corrected.
It happens everywhere though I’m sure.
trafficinsight
Well-Known Member
It occurs to me that in the title of the thread "Experts are lost on how to fix it"
I'mma guess that the experts aren't lost, it's the accountants that are lost.
I'mma guess that the experts aren't lost, it's the accountants that are lost.
Murdoughnut
Well sized member
Found the hacker behind this!
inigo88
Composite-lover
It’s funny because air gapping the INS units from each other and then flagging a disagreement is basically how the old Delco Carousel INS system worked - the ones you’d see on the Concorde, 747 classic, etc:
Carousel Control Display Unit, Commercial Airline
Rectangular display unit, with key pad and a dummy display, to be used to enter waypoints and other navigational information for a long-range commercial airliner.This is the display and keyboard, manufactured by the Delco Division of General Motors, for an Inertial Navigation Sys
I’ve never used them but I’ve watched enough old aviation videos to know that the Orange remote button would link them together for INIT POS, then you’d deselect it and the three INSes would float separately for the duration of the flight (which included long international flights with ocean crossings, etc). I believe there was a comparative logic built it that was smart enough where if one unit drifted enough relative to the other two it would alert and then the remaining two would vote that one off the island.
Any retired JC members get to use these things early in their career?
It’s kind of hilarious that Concordes were crossing the Atlantic at twice the speed of sound with these things in the 70s and now here we are in 2023 lamenting how we’ll ever be able to figure out how an INS/GPS disagree in our subsonic airliners.
The answer is simple - it was arrogant for avionics engineers to consider GPS as an infallible truth source. It’s not. Spoofing works by the gradual introduction of error and allowing the systems to update the INS position multiple times… it shouldn’t work that way. Original calculated INS position should continue to be tracked in memory for the duration of the flight and air gapped from each other, just as it was with the old Delco units. Rather than a truth source, GPS becomes a 4th “potential” position source weighted against the others. And since laser ring gyros are pretty darn accurate one can predict how much their maximum allowable drift should be over the duration of your flight - then your computer could cross-check that against GPS. Has the GPS position drifted more than the allowable INS drift? Then you are being spoofed, and your GPS data is bogus - vote kick it and revert to averaged INS position using the pre-corrected values stored in memory.
We definitely have the tech to do this, the current systems are just unequally weighing the inputs. Add in a 5th ground based nav input with DME/DME and we really have no excuse.
ppragman
FLIPY FLAPS!
You can start doing statistics on this too; it's kind of a solved problem already, to be honest. Just slap a Kalman filter on the whole thing and your final position estimate would be "not that bad" even with a spoofed GPS... now your state vector is being updated by yet another source of truth.
With the data we're already getting you could start to build "probability of being spoofed" into the model with some spatial stats. In places where spoofing hasn't been reported, the probability of spoofing is lower and GPS is weighted as "higher likelihood of trust" or whatever. That wouldn't be perfect, but that would be better than what we have right now, which is basically, "trust your GPS and, uhhh, I guess be careful?" You could even build models of "probability of failure" for various units over time, or "probability of having been spoofed" or something, so that eventually you have a mathematical model where the internal logic can say, "oh, we're being spoofed! ok, well, ignore GPS input for a bit" but you're spot on considering GPS an infallible source of truth is a bigtime institutional failure.
The reliance on it is kind of goofy. Wasn't there an issue a few years ago with Phenom 300s (any Phenom guys here?) where the yaw damp wouldn't work without the GPS or something?
Personally, I'd like to see some alternative navigation tools. Computer Vision for stellar nav is something I'd like to figure out now that I've been in this space for a year - that or something similar.
BEEF SUPREME
Well-Known Member
I knew learning celestial navigation was going to pay off one day.
ppragman
FLIPY FLAPS!
Did you actually learn? I've always wanted to be good enough at that to actually be able to... you know, actually do it and not just be conversant in the theory. There's a badass paper out of some random place in vietnam that had a cool (and much easier) way of doing it with the azimuth of the stars that I wanted to try to implement:
BEEF SUPREME
Well-Known Member
Yeah during another lifetime when I crewed on tall ships. I did a lot of bluewater sailing. I basically taught myself. It's not super difficult. I could shoot the stars, sun and planets but was not sharp enough to figure out the moon.
Usually I was within 3-5 miles of the GPS.
ppragman
FLIPY FLAPS!
That's so rad. I should look into getting sharp at that, I've got a sextant I bought when we lived in Hawaii, but I never really got good at using it or managed to really ever figure out how to navigate with it.
Lunar distance is the method to figure out position without a clock (basically using the moon as a clock?), right? Anyway, yah, this makes me smile, I need to pick up that book again.
