Ground Collision Avoidance System ‘Saves’ First F-16 in Syria | Hacker News Reader

Ground Collision Avoidance System ‘Saves’ First F-16 in Syria

aviationweek.com

130 pointsaerocapture60 comments

The USAF finally went for that? It was first demonstrated in 1998, and has been used by the Swedish air force for years. Their slogan is "You can't fly any lower". Here's the 1998 writeup, copied from Aviation Leak: http://www.f-16.net/f-16_versions_article8.html

The technical paper: http://www.icas.org/ICAS_ARCHIVE/ICAS1998/PAPERS/182.PDF

The actions this system takes are drastic. Roll rates to 180 degrees/sec to get to wings-level, then a 5G pull-up. The pilot's helmet may be banged against the canopy. It's so drastic because flying 150 feet off the ground in mountainous terrain is normal procedure for fighters. If the system has to act to avoid a collision, that action has to be very aggressive.

Here's what it looks like to a pilot:

https://www.youtube.com/watch?v=aPr2LWctwYQ

Here, the pilot puts the plane into an insane bank and, as he says, "goes to sleep" and releases the controls.

... and here is a longer video from NASA about their development of collision avoidance system that can work for regular planes too: https://www.youtube.com/watch?v=XWx6-aK8Ick

That's so NASA PR. About 80% of the stuff in that video was not done by NASA. The F-16 auto-GCAS was a USAF/Swedish Air Force/Lockheed project. The NASA people are just reusing that technology on a model aircraft.

Reminds me of a description of air combat where the pursued would dive, and then pull out at the last possible moment, hoping the pursuer would misjudge and pull out too late. One pulled out so low he raised dust on the ground.

My father liked to attack ack-ack positions by diving vertically on them, as the gun crew obviously was reluctant to fire straight up. Of course, you gotta keep a real close eye on your altitude and airspeed doing that.

There was an incident (among many) between a Swedish figher and the Russian figher on Swedish territory maybe 20 years ago where the Russian was following the Swedish fighter. The Swedish planes being more sophisticated then the Russian fighters, the Swedish pilot went for the sea, straight bown. He pulled back when he knew his plane would make it, but the Russian figher jet could not handle the G's, broke and crashed into the sea killing the pilot.

>misjudge and pull out too late

Su-30 at airshow:

https://www.youtube.com/watch?v=YcUvtnL5p4Y

This is really interesting. I found a PDF with some neat details and imagery describing the thinking that goes into Auto GCAS:

http://www.sfte2013.com/files/78993619.pdf

I don't know if it's just coincidental, but a lot of the examples seem to be concerned with keeping jets from flying into mountains, as opposed to flying into level ground, which was the first thing that came into mind when I saw "Ground Collision."

Cool find! The pictures (and one slide) show an Android phone being used on a "Dryden drone", looks like it might have been a prototyping platform. Would love to see more detail about this.

Edit: Ah, here's a senior project at CalPoly that was sponsored by NASA and looks related: http://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?articl...

Quite impressive senior project. I know CalPoly aerospace does well with interacting with the industry and NASA.

I get "This site does not exist within the LDAP database." when I try that link. Mirror?

Here's a recent video from NASA's Armstrong Flight Research Center discussing the Automatic Collision Avoidance Technology and the DROID UAV platform specifically.

https://www.youtube.com/watch?v=XWx6-aK8Ick

"According to the Air Force, 26% of aircraft losses and 75% of all F-16 fatalities are caused by such accidents."

Wow.

Keep in mind that jet fighter crashes are often survived by the pilot if he or she has the time to react and bail out. Many crashes are caused by technical malfunction or loss of control at altitudes where the pilot has plenty of time to react.

However, if the pilot is still on board, ground collision at flight speed is almost universally fatal. Fighter pilots during ground attack or low-level dogfighting are always pushing the envelope both in aerodynamic/structural capability and maneuvering wrt. terrain hazards, so ground collision is a very serious safety hazard. Military pilots have much, much smaller margins of error than commercial or recreational pilots.

Yes, the wow was more about the 26%.

The article specifically talks about controlled flight into terrain, which, by definition excludes malfunctions.

I think the point being made about malfunctions is that they are less likely to lead to fatalities than CFIT. This would explain the high percentage of fatalities caused by CFIT.

The Stuka had a slightly less sophisticated version of this in 1935:

http://en.wikipedia.org/wiki/Junkers_Ju_87#Diving_procedure

Significantly less, but still impressive for its purpose.

Can you imagine the testing rigor you would need for this sort of system, knowing that if you incorrectly engaged or steered the aircraft, you would be responsible for killing the pilot?

My thought exactly. Warning! Warning! NullPointerException!

No, that's a contract I do not want.

I remember reading somewhere that these guys get all the time they want to program these systems, and the environment is very relaxed - I suppose for this very reason.

Write one line of code everyday, make sure it's bug free.

I wouldn't be surprised if they used functional programming and avoiding object orientated programming.

Why wouldn't they be using Ada? That's exactly what is was developed for.[1]

Does anybody know if Ada is still in use for writing new applications?

[1] https://en.wikipedia.org/wiki/Ada_%28programming_language%29

http://search.lockheedmartinjobs.com/ShowJob/Id/25009/Softwa...

You can take it a step further and use SPARK:

http://en.wikipedia.org/wiki/SPARK_%28programming_language%2...

Excellent. I wasn't aware of SPARK.

It could have been useful for Toyota, rather than the ad-hoc mess they wound up with controlling their cars. https://news.ycombinator.com/item?id=8905718

Boeing for one still uses a lot of Ada for critical flight systems. The 787 flight control software for one.

I left the defence industry eight years ago now, but back then (and based on that job posting below it looks like it's possibly still going on), it was a weird combination of ADA, C++ (sometimes C instead) and Java. In the late 90s, there was a big push in both the US and the UK defence industries to start using off-the-shelf software more, which meant moving to C++ from ADA.

But you'd sometimes have weird systems of ADA being used for the "safety-critical" stuff, which interfaced via CORBA with C++ code controlling it (as a bigger system), and would then sometimes be exposed to the user again via CORBA in Java UI code! (never saw this (Java) on aircraft thankfully, only ships)

A certain UK system also had code for controlling the HUD in ADA, but the main flight control system in C++, again via CORBA interfaces..

This wasn't always the case - some Solaris stuff on submarines was all C++ (with Sun's compiler), but it was frightening the complexity of it.

See http://www.fastcompany.com/28121/they-write-right-stuff for similar

The meatware are already notorious for driving the aircraft into the ground. How much worse could the software people make it? Anyway they've had twenty years to work on this system according to the article.

We have programmed such systems.

We have programs to test all the possibilities in the software.

And we have programs to test the testing programs.

All hardware has redundancy. The software is executed in multiple processors at the same time, so if one fails, the others take control.

I've heard (no source sorry) that in early testing in a simulator flying an F-16 over the international date line resulted in a 180 degree roll that was fast enough to render unconscious or kill the pilot, as the software got confused about teh orientation needed to keep the plane flying level.

There was also a flight of F-22s (real, not simulated) that lost all computer and communication systems and had to follow another plane to get home because of a software bug when crossing the international date line: http://www.dailytech.com/Lockheeds+F22+Raptor+Gets+Zapped+by...

I heard a slightly different version of the F-16 going inverted - it was in the flight simulator during initial software checkout, and he flew into the southern hemisphere, when the plane flipped. Obviously a sign problem...

Imagine being the pilot doing the first real-world test. Point your jet fighter directly at the ground and let go of the controls, then trust that the software written by some guy you've never met will automatically avoid the crash for you.

That's great.

Maybe someone here can enlighten me as to why the systems, specifically TCAS and GPWS in modern civilian planes are only ever used to issue warnings/recommendations, but never take control? it would have at least prevented the überlingen mid air collision and probably some other CFIR incidents in the past years.

I thought about this for a while and couldn't come up with a really good reason.

Auto GCAS also used in cases of G-LOC (G-Induced Loss of Consciousness) which is practically unheard of in a civilian airliner that is able to land safely afterwards.

Fighter aircraft frequently engage in maneuvers that are seconds away from hitting the ground (or water) if something goes wrong.

überlingen was a tragic anomaly caused by multiple factors, the pilots knew they were on a collision course but due to the conflict between TCAS and ATC they collided. This possibility has been reduced by issuing a TCAS reversal if the other jet responds differently than expected.

Early TCAS versions did not even provide correctives, so systems that even provide RAs you could use for automatic corrections are relatively new. Give it (a lot of) time.

Might be product liability issues. Software malf may have completely different economic consequences in civilian vs. military environments.

Difficulty of international agreements, and the expense.

Upgrading to fully autonomous systems adds a lot of cost but is only going to make a difference in a small number of incidents (civilian aircraft don't fly as low and fast as fighters, and civilian pilots aren't distracted by being shot at).

It's only a matter of time. My guess is that it'll happen within the next 30 years.

This is fascinating stuff. Does anybody know the whole cost of this system? I only was able to find a pdf and it says that the seed money was $2.5 millions. If anybody is interested here's the link: http://www.nasa.gov/sites/default/files/337112main_Auto-GCAS...

I found this PDF: http://www.globalsecurity.org/military/library/budget/fy2015... that says it cost ~2 million to integrate in 2013.

I just finished the book "Skunk Works: A Personal Memoir of My Years of Lockheed"

http://www.amazon.com/Skunk-Works-Personal-Memoir-Lockheed-e...

It was a great book that almost feels like the pre-scrum manifesto applied to building aircraft.

All right, this has nothing to do with the Northrop/GD F-16.

FTA: "Auto GCAS, which was developed by Lockheed Martin’s Skunk Works"

"a 5g pull"

Yikes!

That's a pretty run-of-the-mill acceleration as far as fighter maneuvers go. Momentary accelerations in excess of 10g are not unheard of and the use of an ejection seat exposes the pilot to 12g or even more.

Yeah but ejecting is often career ending because of those G's.

I was curious how traumatic it can be. Went searching and found this abstract from a paper[1] studying ejections in the German Air Force from 1981-1997:

"[86 ejections from 56 aircraft had an] overall success [(survival)] rate of 97.6%. Of all 85 participants, 12 (14%) were uninjured, 41 (48.2%) were slightly injured, and 30 (35.3%) were severely injured. Typical injuries were those of the spine and lower limbs. The most common severe injury was a vertebral fracture caused by ejection acceleration. This is followed by lower limb injuries received during the parachute landing fall."

[1] http://www.ncbi.nlm.nih.gov/pubmed/10596782

Why are we bombing Syria?

Google 'isis' aalong with beheadings, rape, child slavery, sex slaves, burning people alive, burying people alive etc. Alternatively buy a newspaper from time to time.

Funny that you and many others got angry because I asked the question that you want to no longer have anyone ask.

Is each bomb going to kill ISIS, or might it do more? Then "why are we bombing syria" never stops being a relevant question to ask, and never one that is fully answered until the answer becomes that we are not.

Also there are lots of horrible regimes and atrocities around the world at all times. The reason we are bombing this one likely has nothing to do with any of the above, and so remains a valid question.

Jordan is bombing Isis in Syria.

hmm ... as a pilot and student of WW2 combat, I think this might be ok in peace-time, but definitely a problem in war time.

- to avoid radar, you have to fly low - tree-top high - to save ammo, you have to shoot near ground-targets - can the software be fooled in mountainous terrain? - what about off-field landings? - Japan's most effective bombers were kamikaze, and American pilots also considered ramming other aircraft after their ammo ran out - if the software is wrong, does it roll inverted and pull down at 5G? how do you stop it?

The article says that 1. The pilot is first notified that the system is about to take control. and 2. That the pilot can override the system at any time.

From http://www.aerotechnews.com/laafb/files/publications/ala_100..., it looks like it's designed to avoid "nuisance activations": “As Nils got down low, I could see dirt being kicked up off of the desert floor and no nuisance activations.”

1,600 automatic avoidances in 141 hours of flight. Holy crap, that's a lot of "point it at the ground, hope you don't die". Unless they did something else?

Surely they just set the ground level to be higher?

Did you RTFA? Do you honestly believe these utterly basic questions you've posed after a few seconds of consideration never occurred to any Air Force or Lockheed or NASA personnel involved in building this system? Or are you just cynical enough to believe the Air Force isn't really concerned about the "war time" effectiveness of its _fighter_ planes?

Reply on news.ycombinator.com