26
$\begingroup$

Landing on a carrier seems like a difficult and risky task for military pilots, especially at night when the carrier should keep low profile and is practically unlit.

It seems exactly like a problem which can better be performed by a computer rather than a human being. A machine can take in account exactly the motion of the carrier and other parameters and land the aircraft precisely where it should. The fact that there should be no enemy aircraft nearby also means that the initiative of a pilot is not needed.

The system can:

  • Either be exclusively inside the aircraft (doing image recognition of a carrier and trying to determine where to land; not sure how easy and reliable this can get),

  • Or be presented as a subsystem within the carrier which sends the encrypted information to the authenticated aircraft about the exact position of given points of the carrier and a second subsystem inside the aircraft which relies on this information to make the controlled landing.

Why isn't it handled (yet) by a computer? Are there projects of implementing such system?

$\endgroup$
  • 11
    $\begingroup$ If there were projects working on this, they would almost surely be classified, so that part is unlikely to be able to be answered here. $\endgroup$ – reirab Mar 3 '15 at 20:25
  • 3
    $\begingroup$ You'll always want a manual override, as with most systems in an airplane (especially military). The only way to make sure people know how to land is by regularly landing manual (one of the reasons of regular qualifications). $\endgroup$ – Mast Mar 4 '15 at 8:19
  • 1
    $\begingroup$ @Mast: Why? The military can be quite utilitarian. If the cost of regular manual landing (including human-caused crashes) exceeds the costs of the occasional crash due to a system failure, then the system should be automated. $\endgroup$ – MSalters Mar 4 '15 at 13:24
  • 2
    $\begingroup$ While it may be computer assisted, it may surprise you that modern aircraft carriers still manage their decks by pushing little toy airplanes made of spare parts around on a picture of the deck known as "The Ouija Board". Why? It's flexible and it's reliable. It means you can still perform air operations with battle damage. $\endgroup$ – Schwern Mar 4 '15 at 19:33
  • $\begingroup$ Your premise is couched as an argument from incredulity, and your "especially at night" includes your own uninformed assumptions. Landing on a carrier seems like a difficult and risky task for military pilots, especially at night when the carrier should keep low profile and is practically unlit. It's been done successfully for over 80 years. While hazardous (so what?) the accident rate decreased significantly once the NATOPS methodology entered training and standardization (starting in the 1960's. Design improvements in later generations of aircraft reduced risk/hazard as well. $\endgroup$ – KorvinStarmast Oct 4 '16 at 20:47
32
$\begingroup$

The Navy does have systems that are capable of guiding an airplane in for a landing, and some aircraft can use this system to land completely on the autopilot. However, the system has not always been reliable and still has its limits, and other reasons for not always using this system are similar to those for land-based aircraft.

Further reasons for this technology not being more fully developed yet are probably similar to this question.

A completely automated, unmanned landing on a carrier has been done, but only fairly recently. Although a computerized system is theoretically more capable of performing precise carrier landings, the details are complex. A lot of sensing and control must be worked out. The team behind the X-47B was awarded the Collier Trophy for the advances in aeronautics they achieved.

The sensors and processing that the UAV uses to do this are probably not all standard equipment on naval aircraft. Adding them would create extra expense and weight. If the system requires addition to the carrier as well, there is more added expense. And in a combat situation, do you want to rely on systems that could be jammed or failed? If you are going to land the plane without a pilot, you might as well not have the pilot at all, and this is the route that future unmanned systems are taking, but there is a lot more work to be done before they completely replace human pilots.

$\endgroup$
  • 1
    $\begingroup$ Note that there is a gap between a controlled landing (my question) and the lack of pilot at all (linked question). Of course, we absolutely need a pilot when things go wrong, but probably not much when landing in a non-fight, non-disaster situation. $\endgroup$ – Arseni Mourzenko Mar 3 '15 at 20:32
  • $\begingroup$ @MainMa That would mean the pilots get not much routine on the kind of landing used in fight/disaster situations, right? $\endgroup$ – Volker Siegel Mar 3 '15 at 21:01
  • 1
    $\begingroup$ @VolkerSiegel: not necessarily. In Paris subway on partially automated lines where there are still drivers in the cabin, once per day, every driver should drive the train manually without computer assistance to make sure drivers keep the reflexes and are able to control the train in a case of a disaster. The same may work for the pilots. This being said, landing on a carrier might be so difficult that it requires constant, and not occasional practice. $\endgroup$ – Arseni Mourzenko Mar 3 '15 at 21:19
  • 2
    $\begingroup$ @MainMa: As I understand it, landing on a carrier actually is tricky enough to demand just about as much practice as possible. $\endgroup$ – Nathan Tuggy Mar 4 '15 at 6:38
  • 2
    $\begingroup$ @MSalters But does Moore's law hold true for safety-critical systems as well? Versions of Intel's 80386 were produced up until 2007 for satellite use, more than 20 years after introduction of the version for commercial marketplace. From what I've seen, at least in civilian safety-critical systems, tried and tested beats greatest and latest. $\endgroup$ – AndrejaKo Mar 5 '15 at 13:25
33
$\begingroup$

Computers can, in fact, control the approach of a Super Hornet all the way to touchdown. However, there are limits to the sea state. Gross deviations in glide path resulting from a pitching deck would make a coupled approach impractical.

However, the most compelling reason to hand fly all our approaches is that we fly a $60 million jet. Murphy's Law tells us that the most junior guy, with the least experience behind the boat, is going to get an engine fire, at night, in rough seas. If our fictional JO wasn't proficient hand flying the ball behind the boat this could result in catastrophic consequences including both the loss of aircrew and aircraft. As often as possible, we prepare for the 1 percentile, not the 99th percentile, so when we do get that engine fire at night we get the upgrade OK, and not the cut pass/ ejection.

Plus, its just plain eerie letting the jet fly itself to the boat, and the current system isn't reliable enough that most guys would feel comfortable letting it fly itself on a regular basis.

Edit: Heck, new guys aren't even allowed to perform auto approaches (auto throttle) to the boat for the same reason.

$\endgroup$
  • $\begingroup$ Figured you'd chime in here. Are automated landings done beyond the once or twice so a pilot is familiar with the system? Your reasoning for not doing them certainly seems sound. $\endgroup$ – FreeMan Mar 3 '15 at 22:00
  • $\begingroup$ @FreeMan certain airfields have ALCS systems and aircraft can safely practice coupled approaches. It also has the added benefit of having lower minimums (the ACLS system as a whole, not just when coupled) than the CV-ILS. So it gives the aircrew the added benefit of operating in harsher weather. $\endgroup$ – Rhino Driver Mar 3 '15 at 22:06
12
$\begingroup$

Reliability.

If the computer or the sensors it needs are damaged (in combat or otherwise) or malfunctioning then the human has to bring the plane in. The only way the human is going to be able to do that extremely precise flying is through regular disciplined practice. They can't practice if the computer is doing the landing for them.

Until you can leave the human behind you may as well make use of them.

$\endgroup$
  • $\begingroup$ I already discussed this point in my comment. $\endgroup$ – Arseni Mourzenko Mar 3 '15 at 23:28
  • 5
    $\begingroup$ Until you can leave the human behind you may as well make use of them. +1 for that, made me LOL! $\endgroup$ – FreeMan Mar 4 '15 at 13:15
  • 1
    $\begingroup$ Then again, if the human is damaged, you do want that computer. And it's a lot easier to carry a spare computer on board than a spare human. $\endgroup$ – MSalters Mar 4 '15 at 13:37
  • 2
    $\begingroup$ Thankfully, humans tend to operate better while damaged than computers do. $\endgroup$ – user3305 Mar 5 '15 at 12:15
  • 2
    $\begingroup$ @Snowman: the day a machine has the same incentive as a human to safely land a plane it's flying, despite being itself damaged, is the day the Rise of the Machines begins. Probably also the day it ends. $\endgroup$ – Steve Jessop Jul 10 '15 at 17:02
4
$\begingroup$

I flew aboard the USS Nimitz in the mid-1980's in the A7-E corsair. The A7-E was difficult to land because of its turbofan jet engine. You never wanted to find yourself in close, high, with too much power. Your only option at that point, if you haven't been waved off, is to make a big correction by coming back on the throttle. Your correction will be too great, and then you find yourself low and slow, with no power on the aircraft. The LSO knows all this and so is encouraging you to keep the power on, but then you hit the burble behind the deck and drop out of the sky, most probably when you are coming on with military power. Then there is eternity. The engine takes seconds to spool up as you are sinking lower. The ball has been red for quite a while, and possibly disappeared off the OLS. Now the LSO is not encouraging you anymore, but screaming "Wave off! Wave off!." So better not to be here.

One of my squadron mates was always in the top 8 in grades for landing on the ship. He was good. He taught me 2 things. One, keep power always on the aircraft if it is the A7. Nudge the throttle up. It will want to take you high and so push it down gently with the stick. Fly your approach always just a little fast.

Second, he advised that I use the Link-4A system, or ACLS (Automatic Landing System). This is mentioned in the post by @passel. So the ACLS system was a computer controlled approach. Pilots didn't use it because close in it could do some weird stuff to you, such as make large corrections and scare the flight suit off of you. He used it all the time at night, and confirmed those fears, but he also said you can recover from these sorts of glitches. I think the reluctance of pilots to use it was more about control than anything else. Carrier jet pilots have specific personalities, and the ACLS had a mythology surrounding it.

The other reason it wasn't used was that it was severely degraded in rain because the wavelength of the radar needed to control the aircraft was small enough that it would jump to the rain at times. Not sure what they have today. But like @passel, I was a bit surprised it wasn't mentioned. I swore by it.

There were several issues with taking an ACLS approach. First, if you took it to touchdown you didn't get a grade for your landing, and my goal was always to be in the top 8. But, also there were 2 sorts of ACLS approaches you could request: (1) Mode I all the way to the deck, and (2) Mode II to be dropped from control at 30 seconds to impact, I mean trap. He told me to take the Mode II approach, because it will set you up on glide path and on airspeed, then don't touch anything and you will get an OK 3 wire. If nothing else, make only small corrections and get your OK 3 wire.

I became a dedicated fan of ACLS approaches and marveled at the control inputs as I sat there watching the approach. Kind of like watching a computer play chess. Absolutely amazing. No big corrections and everything done on time and fast. Dropped you at in close all set up with a solid green ball. What was difficult about the approach is that you didn't have the mind-tactile sense of making the control inputs yourself down the pike and it was difficult to stay up with the aircraft. It took a different sort of discipline that had to be practiced. At times it would happen that for some reason or another I might get dropped earlier than 30 seconds, and if my mind wasn't right there I would be playing catch up all the way to the deck. That was often a difficult approach.

$\endgroup$
3
$\begingroup$

There is, in fact, a current Navy project that is aimed at providing automated carrier landing capabilities to aircraft. An overview of this project can be found here and on this PDF (page 24).

It goes by the humorously engineered acronym MAGIC CARPET, which stands for Maritime Augmented Guidance with Integrated Controls for Carrier Approach and Recovery Precision Enabling Technologies.

The goal is to utilize a combination of the flaps, via direct lift control, and the auto-throttle to maintain a constant glide slope.

$\endgroup$
1
$\begingroup$

I'm a little surprised that no one mentioned Link-4A. I was in the Navy in the late 70's and early 80s, and even then, while I only worked on shipboard Link-11 systems, I heard stories of the related Link-4A system which was used for Aircraft control and communication. Link-4A had a mode that could auto land aircraft even back then, so it isn't anything new. From a navy pub at http://firecontrolman.tpub.com/14103/css/14103_66.htm

Automatic Carrier Landing System:
The automatic carrier landing system selects aircraft in the order of priority from the pattern and enters them into the final approach. During the final approach, a precision radar tracks the aircraft. Correct information pertaining to the approach is transmitted to the aircraft’s autopilot. When conditions are unfavorable for a landing, the wave-off control is initiated and the aircraft is guided through a short pattern and the landing approach is repeated.

But it was also my understanding at that time, the pilots did not care for the system at all. No one would really like the controls taken out of their hands for such a critical portion of the flight.

$\endgroup$
-2
$\begingroup$

Computers are excellent at solving problems that are predictable and repeatable. Landing on an aircraft carrier is exactly the opposite of that. Computers can land very accurately on a static runway, but a runway that is moving and bouncing around on waves is an entirely different story, especially when you don't have the goal of broadcasting your position to the world. Keep in mind that we don't even have good self-driving cars, yet. Google probably has the most advanced ones and those are currently capped at 25 mph. Fighters on approach will be doing 6-10 times that speed and moving in three dimensions while attempting an approach to something else that is also moving (somewhat unpredictably) in three dimensions and at speeds likely as fast Google's cars. It's not as easy as it might sound.

While it wouldn't surprise me if systems like this are in the works (it would surprise me more if they were not,) it's unlikely that any such development work would be unclassified. Very few programs designed to add advanced capabilities to military systems are unclassified.

$\endgroup$
  • 3
    $\begingroup$ But is a 100,000 ton aircraft carrier really "bouncing around on waves"? It seems that the motion should be somewhat predictable, especially in the last few seconds before landing - the aircraft carrier has so much mass that it's not going to be rising upwards and then a few seconds later, suddenly dropping away without warning (but if it did, the sensors would know and signal a waveoff). Sensors on the aircraft carrier should be able to predict its movement fairly reliably at least for the next few seconds. $\endgroup$ – Johnny Mar 3 '15 at 21:22
  • 1
    $\begingroup$ @Johnny Carriers are indeed massive, but so are waves. I don't have numbers for the amplitude or predictability of deck oscillations, but I wouldn't expect them to be very small relative to the tolerances needed for a carrier landing. I'd be interested if someone can produce some numbers, though. $\endgroup$ – reirab Mar 3 '15 at 21:29
  • 5
    $\begingroup$ @Johnny we are an all weather fighting force and land in sea states so bad that you can see the screws come out of the water at the back of the carrier. The LSO has to manually control the ball because the IFLOLS is unable to accurately show (more like predict) glidepath. Also, waves can be bad enough that CAT shots must be timed to avoid shooting aircraft into oncoming waves. $\endgroup$ – Rhino Driver Mar 3 '15 at 21:47
  • 2
    $\begingroup$ @reirab No experience with carriers, but I have been aboard oil-tankers which are similarly massive. Believe me. They heave in different patterns, but just as badly as a small boat. The power of the sea is something that must be experienced to be believed. $\endgroup$ – Tonny Mar 3 '15 at 22:07
  • 1
    $\begingroup$ @Tonny Yeah, that's what I would expect. Big boats have more inertia, but the effect on oscillation amplitudes of the arm length between CoG and the edge of the deck shouldn't be underestimated. Even a relatively small change in deck angle can produce a very large displacement at the ends of an 1,100-foot-long deck. $\endgroup$ – reirab Mar 3 '15 at 22:17
-4
$\begingroup$

There are three main reasons: reaction time, sensor inadequacy, and predictive intelligence. Taking them in order...

For certain types of stimulus-response activities a human can respond faster and has a better reaction time than a computer. For example, imagine it takes 50 milliseconds to sense something completely, 100 milliseconds to transmit that information to the computer, and 100 milliseconds for the computer to figure out the right action, and 80 milliseconds to transmit the response to an actuator. That's a total of 330 milliseconds. A human can respond to an event in 200 milliseconds. Note that for some types of actions the computer's sense-and-respond loop could be much slower than the numbers I have given. For example, most artificial intelligence algorithms are complicated and take a lot longer than 100 milliseconds to complete. Some types of optimization algorithms, like dynamic programming have unpredictable running time, so they might finish in say 50 milliseconds with one set of inputs and 1000 milliseconds in another.

The next big problem is that the human eye, which is wired in parallel directly to the brain, is a far better sensor than anything available to an aircraft. It has a wider field of view, higher bandwidth and better focusing and movement ability than any man-made optical system. Currently the sensors on planes are things like inertial measurement units, GPS and flight instruments (altimeter etc). These sensors are not enough to reliably land a plane on flight deck except in very calm conditions.

Finally, landing on a moving flight deck requires predictive intelligence. You basically have to guess where the deck is going to be in the future, combine that with estimated wind gusts, and then calculate in the future motion of the aircraft given a range of factors. Putting all this together is currently way beyond the ability of artificial intelligence systems.

There are actually a lot of things MUCH simpler than landing on a flight deck that we cannot do. For example, if the Air Force could just land UAVs in heavy cross winds that would be a huge win for the military. Currently, if the cross wind component is above a certain amount we cannot even land UAVs because they would crash. This is because the UAV cannot react fast enough to a wind gust the way a real pilot can. We lose hundreds of millions of dollars in UAVs every year because of simple stuff like this, and remember these are manned system that are only semi-automated. Fully automated aircraft don't even exist, because they would crash so much as to be much more costly and dangerous than useful.

$\endgroup$
  • 6
    $\begingroup$ @TylerDurden Your answer overemphasizes the predictive intelligence required. Nobody is guessing their way onto the flight deck. There is a whole slew of information available to the pilot, be it the ball, the carrier itself, the LSO, the GCA controller, the TACAN, the CV-ILS or the ACLS. All glide slope systems on the boat (like the ball) are calibrated to operate correctly with a moving deck. Furthermore your conclusion that its impossible is incorrect because ACL Mode 1 is an automated approach that's been around for a very long time. The Navy has also landed drones on the boat. $\endgroup$ – Rhino Driver Mar 3 '15 at 22:45
  • 7
    $\begingroup$ 100ms, 80ms is about the right latency to send data to the other side of the planet. Sending information to a local computer wouldn't have that kind of latency. $\endgroup$ – Lie Ryan Mar 3 '15 at 22:58
  • 8
    $\begingroup$ Your numbers appear to be completely made up, here's a differe - an off the shelf sensor can easily provide 300Hz update rates, so if you need to look at 5 samples to see a trend, that's around 16msec and with a 115kbs serial interface, that's 1msec to transmit the data to the computer (lets pad it to 10msec). 30 msec gives the 3Ghz CPU about 100M clock ticks to decide what to do, another 10 sec to send to the actuator and you're up to around 65msec. $\endgroup$ – Johnny Mar 3 '15 at 23:12
  • 9
    $\begingroup$ @TylerDurden I'm a carrier qualified Naval Aviator. I've done what you're talking about. $\endgroup$ – Rhino Driver Mar 3 '15 at 23:16
  • 2
    $\begingroup$ Worth noting that avionics generally are miles behind the cutting edge technology which may be the cause of disparity in the numbers here. Cutting edge FPGAs can receive, process and transmit data in latencies measured in nanoseconds. See also space.stackexchange.com/questions/247/…. $\endgroup$ – Chiggs Mar 4 '15 at 12:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.