In the past 20 years the following technologies reached full maturity:

  • Computers now have immense computation power, sufficient to process real-time video with several backup systems for a relatively low cost
  • High-resolution and high-frequency cameras, with better resolution capabilities than the human eye
  • Neural networks capable of recognizing objects and solving certain problems at a better rate than humans. Given an aerial image, a neural network could tell you the position of the plane in less than a second anywhere on the planet, for example.
  • Cheap sensors for pathfinding, such as RFID for close range and GPS/Glonass/Baidu/Galileo for long distances

So... what's currently preventing us from building an airplane that would be as complicated to operate as a car? Surely we have the technology to make an airplane do 99% of the work?

  • $\begingroup$ Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, on Aviation Meta, or in Aviation Chat. Comments continuing discussion may be removed. $\endgroup$
    – Jamiec
    Commented May 23, 2023 at 6:48

10 Answers 10


We do have airplanes that can be flown with very little training. If you cut out all of the details that make airline flight as remarkably save as it is, a non-pilot can be put in the pilot's seat of an airliner with very little training and have a flight where things go fine 99%. Plenty of aircraft have autopilots that work uneventfully 99% of the time, and even some light aircraft have autolanding capabilities for emergencies where the pilot is incapacitated.

But 99% isn't good enough. Sensors fail. Autopilots fail. A pilot needs to be able to do everything the automation can do, just in case the automation fails. They also need to be able to competently monitor the automation and be ready to disengage it if it is acting improperly.

There are, of course, other considerations. Unlike a car, where you only have to worry about traffic in your immediate vicinity, planes have to worry about traffic that may be miles away and in any direction. Finding traffic around you and keeping a mental picture of the situation in three dimensions is not something that comes naturally, and must be learned. Anything other than perfect weather can also make the basic flying task much more difficult.

  • 1
    $\begingroup$ For cruise flight that would be accurate, but when it comes to takeoff and landing I don't think a non-pilot would be able to perform those maneuvers even with a CAT IIIC autoland system. $\endgroup$ Commented May 19, 2023 at 23:12
  • 7
    $\begingroup$ Once you have autopilot, the hard part is setting it up correctly. The wrong setting has resulted in more than one crash. $\endgroup$
    – Therac
    Commented May 19, 2023 at 23:41
  • 9
    $\begingroup$ @SMSvonderTann Not with absolutely zero training, but amateurs learn to takeoff and land in full flight simulators all the time, and it's a pretty quick process. $\endgroup$
    – Chris
    Commented May 20, 2023 at 1:14
  • 3
    $\begingroup$ @Therac less than 1% of the time, though, so the answer's point holds. $\endgroup$
    – phoog
    Commented May 20, 2023 at 9:48
  • 2
    $\begingroup$ To pile on to Therac's comment, a 250 hour C172 commercial pilot isn't going to be able to simply "set up the FMS" of an airliner if he/she has never used one before. It's about a week of classes, and there are some gotchas to learn to avoid. So, there would be plenty of "events" during such a hypothetical flight... $\endgroup$ Commented May 22, 2023 at 1:08

Short Answer: Because it takes the exact same skills to land a fully automated plane in a crosswind when the automation fails as it does to land a manual-only plane in a crosswind.

Actually, it would be harder when it happens because your skills erode if you are accustomed to the plane doing all the work most of the time...

Why do we still not have an airplane that can be operated by a person with very little training?

The longer answer depends on what you consider very little training...

The amount of hours of training and practice to get a drivers license vs a solo endorsement, (or even a private pilot certificate) are really not all that far apart. Consider my personal experience:

  • I soloed after 14 hours of flight instruction flying lightly over about 7 weeks. If you are a quick learner and dedicate your time to it this could be compressed. I would consider that "very little", especially in hindsight.
  • For comparison I couldn't drive a car solo until I had my learners permit for 6 months, had completed a semester long driver's education course, and passed a state driving test. I didn't log driving hours, but I'm sure it was a lot more.

What's currently preventing us from building an airplane that would be as complicated to operate as a car?

Nothing, we have them now. Many, in fact most existing 2 - 6 seat GA airplanes are actually a lot less complicated than a modern car. If you want more automation you gotta pay for it, but even budget GA airplanes can be outfitted with some pretty nice autopilot systems for not a lot of money.

However, it almost seems like you are asking about self-flying, fully automated air taxis. Those are in work, but if you want to just push a button and have it take you somewhere go you’d be a passenger, not a pilot. No training required if the automation is reliable enough to be certified for flight without a pilot.

So, do you want to actually fly the airplane, or just go along for a ride?

  • $\begingroup$ For an example, in NY state, someone with a learner’s permit needs 50 hours of supervised driving practice before they can get a full driver’s license. That’s a lot more than the 14 hours you cite. $\endgroup$
    – KRyan
    Commented May 22, 2023 at 18:53
  • $\begingroup$ @KRyan, yeah, it isn't a perfect comparison. 40 hours min is required for FAA Private Cert so you can fly not just solo, but carry passengers. In Washington State, if under 18 I think you can't carry passengers other than immediate family even once you have a Driver's license. Anyway, they are not the same, but comparable in terms of training and experience required, which was my point. $\endgroup$ Commented May 22, 2023 at 19:13
  • $\begingroup$ "it would be harder when it happens because your skills erode if you are accustomed to the plane doing all the work most of the time" - also a huge, lurking problem with self-driving cars $\endgroup$
    – bta
    Commented May 23, 2023 at 1:04

You can fly with very little training, or you can fly with very few accidents.

Small quadcopter drones require almost no training, so that's one solution. Their operation isn't safe, but they're cheap enough to replace when crashed and light enough to not do much damage to people or other property when they crash. Cars do require training to drive, but the public demand for cars outweighs the public demand for safety, so the training is kept accessible to everyone.

Even when built to aviation standards, some of the millions of parts comprising a large aircraft will occasionally fail, compromising one or more systems. No instrument is perfectly reliable. While engineers do implement safeguards and redundancies, they still can't account for every possible combination of failures. That's where humans still have the advantage.

As for neural networks, they are less reliable than aviation-grade instruments. They aren't free from errors that humans can make, and also add their own failure modes like hallucinations (which have made it to SE). They can tell you a position anywhere on the planet, but they can tell it completely wrong if they misclassify an object.

GPS and flight automation are used extensively to navigate in IMC. Immensely useful as they are, they are also prone to fail catastrophically in the event of a sensor glitch, as seen in certain B-2 and 737MAX crashes.

Experienced human pilots are capable of spotting failure modes in flight automation systems, and addressing them in survivable ways. Not every time, but that's how commercial aviation got to the safety record it has now.

Currently, the consensus is that it takes two experienced pilots to fly an airplane safely. The same objections that apply to single pilot operation, apply to the idea of minimally-trained pilots.

  • 5
    $\begingroup$ "Their operation isn't safe, but they're cheap enough to replace when crashed": and small enough that the potential for damage to other property and injury to people is also small. $\endgroup$
    – phoog
    Commented May 20, 2023 at 9:50
  • $\begingroup$ "since they're trained by humans" they are trained by data. The data might come from humans, it might not. You could make an AI that plays chess without using humans at all, just give it the rules and let it play a trillion games on its own to generate the training data. Chess is a great example because it is very much not human biased, chess AI's play chess very differently from how humans play it. On the other hand LLM's sound human because they are trained on data generated by humans (that's the whole point). car AI's are usually a mix of both $\endgroup$
    – eps
    Commented May 22, 2023 at 22:30
  • $\begingroup$ @eps True, I edited to specify more accurate detail about computer vision glitches. $\endgroup$
    – Therac
    Commented May 23, 2023 at 8:47

A car and an airplane do not compare

Contemplate the difficulty of manual control, purely as a baseline. And we're talking in the framework of automation which has failed, where the rider did not expect to be hand-driving/flying this vehicle today, and their task is to reach a place of safety or perhaps "limp home" via wisely chosen routes (e.g. minor back roads).

A car is an on/off button (these days), a forward/reverse selector, a big wheel to aim it, and two pedals for slow and fast. And that's it. That's a pretty obvious control scheme. An untrained person can sort it out fast. Then, all you need is a strong aversion to hitting things, and a willingness to observe and learn, especially what other cars do, and there are plenty around to observe.

OK, so you figured out the "start" and "go" levers in your airplane. Where's "stop"? Found it, is it all right to just alternate between "go" and "stop", no worries about anything overheating, right? We're toddling around the taxiways, anything special we need to do around that BIG taxiway there?

OK we're taking off, when do we rotate? How hard to pull back? Is there a "too hard"? Now we just use the stick to aim the nose where we want to go, right, and the airplane just sorts it out?

See, I'm pointing out here that the baseline tasks of manual flying are much, much more intricate than the baseline tasks of manual driving. And that's going to rear its ugly head as a BIG problem when the automation fails and a non-qualified person needs to get the vehicle to a safe stopping point. Not really a problem for a non-qualified driver to limp a car to a shoulder. Airplane not so much - too many fatal mistakes are possible, and they can happen very quickly - too quick to call 911 to ask how to fly a plane.

"But what about all those videos showing non-pilots getting talked down by ATC?" Caution: Survivor bias. The savior is already in the copilot's seat, has been watching the pilot's moves, may have been given a few quick lessons or a few minutes hand flying (the pilot choosing them for that seat, and educating them, for just this situation)... already has a headset, correct frequency is already tuned in, and from watching the pilot they can find "Push to Talk". And they are flying in a world sparse enough that human ATC helping human pilots is the norm. Now, OP's self-flying airplane won't be the only one. Put 50,000 self-flying planes in that airspace and think about how ATC works now. Any humans are saturated by complaints about their FlyUber having no Diet Coke in the minibar. Or you could have a common-mode failure where thousands of self-fliers revert to manual at the same time and ATC is simply overwhelmed. That "talk them down" thing isn't gonna happen.

In fairness, automation is actually easier in some ways. Look at what is holding back "self-driving" from being properly called that -- it's all the weird chaotic stuff that happens from sharing uncontrolled space. Airports are much relieved of that because of access control. Airports don't have pedestrian crosswalks. A beach ball isn't going to suddenly bounce across the taxiway with an inevitable small child in hot pursuit. So at least for that part, we just need a special Bot-plane-only airport. Unfortunately, real estate developers are not begging the aviation community to take subdivision projects off their hands and turn them into conveniently located airports! So your new robo-airport will probably be 50 miles out unless you luck onto a long, narrow Superfund site - not convenient at all, ask Montreal if that spells success.

But just because the automation is easier, doesn't make it easy.

enter image description here First self-driving "car" wreck. Computer "driver" but human "watcher" - after thousands of hours of watching the computer never make a mistake, the human was complacent or stunned to see the impossible. E-devices were stowed; all indication is the human was as attentive as humans get. Source: NTSB

Humans are good doers, but bad watchers.

So of course you're dying to know: What went wrong in the above photo? Yes, train control lets you overrun the distance you can see, but in this case the sightlines were good enough that an e-stop on sighting would've sufficed. This wound up being a harbinger. The problem is those pesky human factors... artfully described in the movie Sully (setting aside its accuracy, if any).

Remember - the same tech that is "improving" aviation is improving cars too - we've been working on self-driving cars for some years now, and how's that going?

The pattern seems to be that computer mistakes are different - they have no sense of scale, and so their mistakes can be whoppers. They will drive into the side of a turning semi, or disregard a pedestrian, because the thing is below detection threshold or not expected. Attentive humans do make mistakes, but they don't make large blunders like this as a rule. And so the "solution" is thought to be human watchers...

But it turns out, humans are good "doers" but really bad "watchers". The task simply does not suit our brains. The Washington Metro rail accident taught us this, but we seem doomed to re-learn it. Many self-driving regimes are giving humans the job of watching for those "once every 100 hours" edge conditions and jumping in immediately. This does not work. Actually, the human has a cognitive crash - it takes time to shift out of complacency, realize an edge condition is happening, that there's a call for action right now, and shift from passive "watcher" to active "doer". Why stop in civil transport? Take the Moskva - its clunky Soviet systems needed an "unblinking eye" from its operators, and didn't provide strong automation assistance like Patriot does. Aside from organic human complacency, you also have the impulse for electronic distraction - present in several "self-driving" car wrecks and allegedly, one commercial flight overflying its destination by 20 minutes.

On the other hand, when you flip the script and let the human be the "doer" and the computer be the "unblinking eye" of a watcher, that combination works out pretty well. It works better for the computer to say "terrain, gear" (not knowing if you're trying to land) than for the human to know the computer has latched onto too low a glideslope (a-la Die Hard II, but that actually happened).

The failure modes are much, much worse.

We already have some data to know what "self-flying airplane" failures look like. Like the odd turn of weather that Air France 447 flew into that froze their pitot tubes, necessitating a very easy, by aviation standards response of "just leave throttle and trims where they are, since they are correct, and don't do anything crazy with the stick". Or the "intermittent runaway trim" problems on several 737s that flummoxed four trained pilots who train runaway trim scenarios in the simulator.

Of course there are countless stories of the automation doing something wacky and the pilots saving the plane just fine. But I simply do not see how an untrained non-pilot could possibly sort out this kind of computer failure that already does happen to trained pilots. What happens then, when all pitot tubes freeze up due to odd weather, or whatever other automation failure renders the autopilot unable to fly? Do we just accept that the plane crashes?

Or do we try to get the poor passenger to save the day -- to which I say how? I mean, normal airplanes have all the kit in place - stick, instruments, throttle. There's a radio system and headsets already tuned to talk to ATC. There are many cases of "pilot incapacitated" situations where passengers were able to don headsets, talk to ATC and get the plane down intact. I'd expect those to be eliminated for cost in a self-flyer. Now what if there's no ATC to talk to? What if there's no headset? What if there's no stick?

OK, so you provide all those things, how do you keep the untrained passenger from using them inappropriately? They have to understand the fundamentals of flight, such as "there is a speed called 'too slow' and you need to know what a 'stall' is and how to get out of it". Wind shear is a thing. If they don't have stall-spin recovery training, they probably will have a bad day, especially if they are in IMC.

You could set it up "fly by wire" so the computer disables them unless it's in trouble, but can we count on the computer to know it's in trouble? In the WMATA crash and the self-driving accidents I mentioned, the computer thought everything was hunky-dory.

The premise of this question seems to be **the automation does not fail; or it fails, you die, and that's just accepted as the cost of flight.

The problem you have then is people on the ground. There's an established FAA gold standard for how good an airplane has to be, to be allowed to fly over populated areas. The self-flying plane would have to meet that.

A new hope

I'm thinking I should mention an X-factor here: drone deliveries. So far, they're not used for much except delivering ordnance to soldiers on front lines. But if "drone delivery to homes" starts happening in earnest, we will begin to collect large volumes of data about reliability and failure modes.

As this corpus of hard data is developed, it will give us a new view to the question of putting humans in such vehicles. This may invalidate the above factors.

But here, there'll be no substitute for vast experience. They'll have to show millions of incident-free deliveries*, or show how the accidents which do happen are survivable.

* Military deliveries notwithstanding, it turns out some soldiers are revanchist and are not so happy to receive deliveries via this new mode, even though delivery is free.

  • 6
    $\begingroup$ I suspect that the perceived difficulty of driving a car is in fact cultural, mostly attributable to the different rigor required in the training of new drivers. Germany is a far cry from the US in this regard, for example. $\endgroup$
    – phoog
    Commented May 20, 2023 at 10:42
  • 6
    $\begingroup$ @JonathanReez For cars, it has to be better than the average driver. For planes, it has to be better than the average pilot. Most pilots who aren't out for the joy of flying are professionals, so the standard is much higher. I also believe that self driving cars have the advantage of having rapidly adjustable equipment and rapidly adapting situations, where as planes take forever to accelerate or turn and by design are never in situations where sub-second reaction times help. $\endgroup$
    – prosfilaes
    Commented May 20, 2023 at 15:04
  • 4
    $\begingroup$ ? There are no self-driving cars at all. (You could possibly argue there are some Level3/4 cars, but they're only usable in extremely limited, very slow, settings.) faistgroup.com/news/autonomous-vehicles-levels $\endgroup$
    – Fattie
    Commented May 20, 2023 at 15:50
  • 6
    $\begingroup$ @JonathanReez Even prosfilaes is understating it - you don't need to be as reliable as pilots, you need to be as reliable as Part 121 aviation - the commercial stuff, which sets an absolutely bonkers bar for safety. That is because citizens are irrational toward flying safety, and demand a much higher safety bar than statistics alone call for. And they are irrational again about self-driving anything. Have you ever seen me rant on DIY stackexhange about how humans are terrible at risk assessment and do it emotionally rather than scientifically? That also. $\endgroup$ Commented May 20, 2023 at 19:16
  • 3
    $\begingroup$ @Harper-ReinstateMonica I'd argue the safety bar for flying is reasonable; the FAA even does a cost-benefit analysis for major rules. Airlines are below trains and buses per passenger-mile, but about the same per journey. It's only private cars that stand out. People are more tolerant of car crashes due to the superiority bias. $\endgroup$
    – Therac
    Commented May 21, 2023 at 8:25

One aspect is WHY people fly GA aircraft. Light aircraft are a pretty silly way to travel, if just travel is the goal. Apart from anything else they generally do not have a toilet!

So for most people flying in that size aircraft the reason is because they WANT the experience of flying themselves, or they are doing something out of the ordinary where flying is the only option (bush flying, law enforcement/emergency services).

Aircraft like the Cirrus SR22 are poking this space from one end in terms of assistive smarts, and the emergency service end is upgrading to allow already skilled pilots to operate into more extreme weather/environments.

Merging these two would get you a light aircraft that in good weather could fly itself around specially configured airports (since ATC needs to be able to talk to your autopilot) and might even be allowed cross country on pre approved routes but would not expect them to be allowed near/over urban areas or into actually remote regions.

Neither of these market sectors are after something so limiting so the already small GA manufacturing market has not spent the money to certify the hardware then jump through the relevant regulatory hurdles.

Further narrowing the pool of potential buyers would be that regulatory approval for flying without a certified pilot onboard almost certainly involves ATC having a kill switch (hopefully involving a parachute!) to prevent these vehicles becoming a hazard to others.

  • 2
    $\begingroup$ "The SR22 series has been the world's best-selling general aviation (GA) airplane every year since 2003" ... I had absolutely no idea that was the case! $\endgroup$
    – Fattie
    Commented May 20, 2023 at 15:29
  • 1
    $\begingroup$ Amazing reading .. cirruspilots.org/Safety/CAPS-Event-History $\endgroup$
    – Fattie
    Commented May 20, 2023 at 15:39
  • $\begingroup$ Under en.wikipedia.org/wiki/Cirrus_SR22#Safety_record it is relevant to note: By the end of 2013, the accident rate had been reduced to a fatality rate of 1.01 per 100,000 flight hours. This was attributed to better training, particularly in use of the ballistic parachute system., in particular "better training". $\endgroup$ Commented May 22, 2023 at 11:12

I think the big reason why this hasn't been developed is money.

How much would it cost to develop this sort of airplane? We'd basically be developing an autonomous aircraft that can fly from one civilian airport to another using the same airspace as other VFR and IFR flights, and handle emergency situations on its own. We'd also have to do extensive testing in order to make sure that it's safe. I think that most of the necessary technology probably exists already, but all that development and testing would be a tremendous amount of work.

This technology wouldn't make flying very much cheaper, because most of the cost of flying goes towards the airplanes, not towards the pilots. So I think it would be very hard for a private company to recoup the cost of developing this technology.

  • $\begingroup$ In a sense this is the right answer. There's simply no real demand for it - why? i honestly don't see that much demand for self-driving personal cars - far less personal planes. It's interesting that self-driving trains-monorail type transport, is pretty easy to do - but it's very little used, there's just no great need. $\endgroup$
    – Fattie
    Commented May 20, 2023 at 15:42
  • 3
    $\begingroup$ This might be on the right track, but I’m feeling like the fact that safe, reliable, usable self-driving cars don’t already exist, despite high levels of demand and investment, suggests that greater aircraft automation with acceptable levels of safety and utility is also still out of reach. $\endgroup$ Commented May 20, 2023 at 19:52
  • $\begingroup$ There are in fact companies developing this technology today. See Wisk, Xwing, Joby. forbes.com/sites/suzannerowankelleher/2023/02/26/… $\endgroup$
    – AShelly
    Commented May 21, 2023 at 6:39
  • 1
    $\begingroup$ I think your estimate of what commercial pilots charge is on the low side, by at least an order of magnitude. Each flight needs at least two of them, and it's one of the higher paid jobs around. They also represent at least two seats which could be sold - seats with a view out the front window might even attract a premium. $\endgroup$
    – craq
    Commented May 22, 2023 at 3:10
  • 1
    $\begingroup$ Charging per flight isn't the only possible business model. Removing the flight deck, all of the pilot equipment, and never having to worry about pilot schedules will easily be worth millions per plane upfront, plus a couple million annually. $\endgroup$
    – Therac
    Commented May 22, 2023 at 8:04

Here in Austria you can become a glider pilot at 16 years¹. You only need 15 hours of flying time (10 of those with teacher) and 45 landings. For motor gliders you need an additional 20 landings and 10 hours of flying time.

Compare that to a driver’s license: You need 18 “sessions” of training (or only 11 but then have to drive ≥1000km with a supervisor e.g. your parents) plus first aid course. Then after you have the license you need 4 additional training sessions and have to take a safe driving course.

I’ve only flown a motor glider for an hour with a teacher out of curiosity, so I’m far from experienced. I don’t see how modern technology (apart from a GNSS device) could make flying a small plane any easier unless it’s fully automated and you basically just enter your destination airport. Flying is inherently three-dimensional which makes it more complex. But in my experience it’s really not that different from flying an oldschool RC plane or helicopter (those are actually damn hard to fly!).

From what I’ve seen of larger aircraft (e.g. Boeing 737) they do seem to have an obscene amount of controls and I bet a good user interface designer could optimize this a lot without compromising reliability or features in any way.

¹ training can start at any age, but you can only get the license at your 16th birthday or later

  • 1
    $\begingroup$ It might be worth re-reading the question and in particular the last line of the about 'having the aircraft do 99% of the work', and see if you can edit your question for how your motor glider experince would have been different if 99% of the work had been done by the glide. This is a good answer, just for a slightly different question. Re the fourth para graph recommend a read of aviation.stackexchange.com/q/2873/60886 $\endgroup$ Commented May 20, 2023 at 6:43
  • $\begingroup$ I bet a good user interface designer could optimize this a lot without compromising reliability or features in any way Only to a very limited degree. A good part of the best (== safest) UI for critical systems is quite the opposite of what works well for low-level, non-critical systems. Multiple menus to get to key controls? Not good in an emergency. All switches look and feel the same? Not good in an emergency - you want different key pieces to look and feel different so that you are less likely to accidentally hit the wrong switch. Hide the circuit breakers? Not good in an emergency $\endgroup$ Commented May 22, 2023 at 16:49
  • $\begingroup$ where you need to turn off a system that is causing problems (interference, fire, etc.) Plus a lot of stuff that is now on the main "glass cockpit" screens needs separate backup displays for the 0.01% of the time when that main screen is unavailable but you still need to fly the plane. A car can bend more of these "rules" because even at 75 MPH you can usually pull over to a safe stop in under a minute and figure it out (reboot or wait for AAA, etc.) safely at the side of the road - that doesn't work in an emergency at 30,000 feet. $\endgroup$ Commented May 22, 2023 at 16:51

Many good answers with a lot of different perspective. I'll tackle a very peculiar and abstract aspect: potential damage.

Harper already mentioned in his answer that the failure modes are much harder. I want to focus on that specifically from a basic physics POV.

When you crash with a vehicle (i.e. come to an abrupt stop) what is causing the damage, both to the vehicle and to the obstacle, is the mechanical energy dissipated in the collision (let's neglect for a moment possible fires and explosions due to the fuel).

For a car the energy is essentially kinetic, since the potential gravitational energy is usually not involved (the crashed car usually remains at the same distance from ground, unless it falls down a cliff or something like that).

A plane moves in a further dimension and by flying higher it increases its potential energy. When crashing to ground the potential energy that was accumulated by flying up will be also dissipated (technically it would be converted to vertical kinetic energy during the fall).

Keeping in mind these two formulas:

  • Kinetic energy: $ E_k = \frac 1 2 m v^2 $
  • Potential energy: $ E_p = m g h $

where $m$ is the mass of the vehicle, $v$ its horizontal speed (we will assume the plane is flying horizontally before beginning to fall), $h$ is the flying height (distance from ground) and $g$ is the gravity acceleration ($9.8m/s^2$).

let's put some numbers into this problem and let's compare a car versus a small plane of the same mass and flying with a speed equal to the speed of the car:

  • v = 100 km/h = 28 m/s
  • m = 2000 kg
  • h = 1000 m

We get for the kinetic energy of both vehicles:

$$ E_k = \frac 1 2 \times 2000 kg \times (28 m/s)^2 = 784 kJ $$

Whereas for the potential energy of the plane: $$ E_p = m g h = 2000 kg \times 9.8 m/s^2 \times 1000 = 19.6 MJ $$

So the plane has a surplus of "destructive energy" that amounts of its potential energy, and that energy surplus is way (more than 10 times) higher than the "destructive energy" of a car.

So the damage it could do when crashing will be more than 10 times worse (in this extremely "spherical cow" simplified scenario).

Therefore the risk of allowing an untrained person to pilot a plane when something goes wrong in the automated system is literally orders of magnitude higher.

This is something probably any sane legislator would prohibit, even if someone would be willing for paying for its costs (huge risk/benefit ratio).


Thanks to Peter Cordes for spotting a dumb calculation mistake in the end conclusions (I should avoid posting late at night :-).

He pointed out a couple other issues with my analysis, with which I don't agree. Not because they are not relevant, but because they are well out of scope of this answer.

I didn't want to make an accurate risk analysis of a real scenario (other people answered in that direction already, and I'm not really an expert of aviation risk analysis), I just wanted to take another, more abstract, route using very basic physics to show that what the OP envisions has an intrinsic order of magnitude increase in damage potential.

Of course the actual risk assessment should be done on more realistic scenario, including the average weight of an aircraft that could realistically be fitted with such an advanced system (I used the same weight of a car, which for a plane corresponds probably to a bottom of the line small aircraft), danger coming from the residual fuel ignition, air drag and so on.

Moreover, note that I focused on the damage potential, not on the actual risk of an accident happening. Of course in a detailed analysis the estimated risk should be compounded with the potential damage. My gut feeling is that overall the result would be worse (i.e. there would be even less incentive to let untrained persons pilot "automated planes"), but I could be proven wrong by a better real-world analysis

  • $\begingroup$ With the exception of intentional acts such as 9/11, when was the last time a plane crash caused a lot of damage on the ground? $\endgroup$ Commented May 21, 2023 at 18:56
  • 1
    $\begingroup$ @JonathanReez You didn't ask for records, but for reasons why one wouldn't want untrained pilots piloting "autonomous" crafts. You made the comparisons with cars. Each years tens of thousands of "untrained" drivers crash into something. This is a risk assessment analysis: the potential damage should be compounded with the enhanced risk of an untrained pilot to not recover from a system failure. Moreover, avoiding a costly and difficult training would allow more people to fly, increasing the risk. $\endgroup$ Commented May 21, 2023 at 20:11
  • $\begingroup$ You're ignoring air drag during descent from 1000 m altitude. A lot of that potential energy is dissipated in the atmosphere, even in a worst-case failure mode where the AI pilot goes into a vertical nose-dive (and the speed doesn't tear the plane apart into multiple lighter pieces before impact). More realistically, light aircraft speeds could be maybe 2 to 4x highway speeds. Like 200 knots is 370.4 km/hour, and that's fast for a light aircraft (Cessna 172 V_ne is 182 MPH). So 4x to 16x the kinetic energy for the same mass, over a wider area (wings instead of just the car's nose). $\endgroup$ Commented May 22, 2023 at 1:55
  • $\begingroup$ (Another light aircraft mentioned in another answer, cirrusaircraft.com/aircraft/sr22t SR22T has a Vne of 205 kts indicated. (Its max cruise of 213K TAS is at higher altitude; the max normal-operation IAS is 176 kts). So we're still in the ballpark of 10x the energy of a car crash, not 10000x. Obviously a vertical dive ground could exceed Vne, and the wings probably wouldn't rip off or the structure break up until some higher speed, especially if an autopilot isn't even trying to pull out of the dive. $\endgroup$ Commented May 22, 2023 at 2:11
  • 1
    $\begingroup$ Wait a minute, 19.6 MJ is 25x the energy of 784kJ, not 10000x. 19.6 / 0.784 = 25. So that corresponds to an impact speed of 5x as fast as a car at highway speed. (And is the speed you'd reach falling from rest from 1000 m without air resistance.) So that's 500 km/h = 269 knots, barely plausible as a worst case for a light plane. (At sea level, true air speed is about the same as indicated, and we can assume vertical wind speed is negligible. Crashes at higher-altitude terrain have lower IAS for the same impact energy, less drag.) At sea level, 500 km/h is Mach 0.4 $\endgroup$ Commented May 22, 2023 at 2:15

A couple of issues I haven't yet seen mentioned:

  1. If the entity (automated or human) piloting an autonomous land-based vehicle develops a fault that would preclude continued safe operation, the vehicle can generally be brought to a safe stop fairly quickly. Airborne vehicles, however, often cannot land safely without minutes' if not hours' notice. Thus, a fully-autonomous air vehicle would need a level of safety and redundancy far beyond what would be practical for a private vehicle.

  2. The number of airplanes presently in the sky allows enough distance between them that a pilot would generally have many seconds advance notice of any potential collisions, which is important because it will generally take multiple seconds for an airplane to significantly alter its trajectory, or to re-establish a desired trajectory if it is disrupted by turbulence. Even if one were to try to divide the airspace into a precise set of paths in 3d space, the number of aircraft that could be safely airborne over a square mile of land area would still be a small fraction of the number of cars that can occupy roads on a square mile of land.

If the cost of an autonomous flying vehicle were to fall too much below the cost of using a conventional aircraft and hiring a pilot, it would be necessary to impose taxes and fees to limit the number of flying vehicles, making it impossible for flying in an autonomous vehicle to be much cheaper than flying in a piloted one.

  • $\begingroup$ There’s always a safe (for others) way to land a plane. Just crash it into the closest empty spot of land and you’re done. Even if you’re flying over NYC such spots are at most 5 minutes of flight away. The operators of such flights would assume all risk and that’s it. $\endgroup$ Commented May 22, 2023 at 15:23
  • 1
    $\begingroup$ @JonathanReez: Designing an air-traffic-control system to accommodate the possibility that any aircraft might need to make a no-notice emergency descent would severely limit the achievable traffic density, and even then the time required to make a "safe: emergency landing in an aircraft would, for any level of "safety", still be at least an order of magnitude longer than would be needed in a road vehicle. $\endgroup$
    – supercat
    Commented May 22, 2023 at 15:49
  • $\begingroup$ @JonathanReez: You're assuming a level of control that is not "always" available (especially because contemplating crashes, we are talking about emergency situations in the first place). $\endgroup$
    – Ben Voigt
    Commented May 22, 2023 at 16:53

Yes, there are airplanes that, from the regulations point of view, can be flown with very little trainig: the starting ultralight class, with weight and performance limitations, does not require a pilot license, nor an aircraft registration. You build it, jump inside, and are allowed flying. You better read about mechanics of flight and basic aircraft handling before this, and fly only in days of 'sun and flies' Blessings +


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .