I know that we have a number of professional pilot members at MUDSPIKE, and to that point I want to be clear that this discussion is meant to be neither a resentful jab nor troll. Rather, given the current state of drone and autonomous system development, I think that the discussion is relevant, especially considering at some point those of us with children will be guiding our offspring either toward or away from aviation.
And as soon as I make a sweeping opinion on what the near future holds, an entity like Google, Tesla, or SpaceX decidedly proves me wrong.
For those of you well into your aviation careers, perhaps this discussion is mostly irrelevant. Iāve read various schools of thought concerning whether the F-35 is that last manned fighter/attack aircraft purchased by the military.
Regardless, I found this AVWeb article an interesting read on the subject, and didnāt exactly contain the responses that I would have expected from those surveyed, EAA Chairman Jack Peltonās for instance.
I understand all of the economic and safety arguments, but I canāt feel like the discussion ignores the reason that a lot of us decided to climb into cockpits in the first place. Hopefully there is a spot reserved for the pilot enthusiast, in other words, someone not wholly focused on either trying to get from point A to point B or putting ordinance on target.
Iāll tell you the one reason Iām not worried: irrational fear.
Give the world an automated airliner that can perform 999,999 hours of safe flying, but it suffers one crash in that millionth hour. The outcry will be swift and irrational. Even if you present data that says human operated airliners will crash four times over the same time period. (These are not real numbers, Iām just making a for-instance scenarioā¦)
I still wonder if even automated cars will make it in the long run. Despite the math that might eventually say they are 50x or 100x safer than human driven cars, that one computer driven car accident caught on video will sort of taint the well. People are distrustful of thingsā¦even in the face of facts.
āBehind the curtain, aircraft manufacturers are working on a single-pilot cockpit where the airplane can be controlled from the ground and only in case of malfunction does the pilot of the plane interfere. Basically the flight will be autonomous and I expect this to happen in the next five to six years for freighters.ā
Iām not sure how often this guy has dealt with government agencies when he proposes half a million pound aircraft flying around via computer in five or six years. It will take five or six years to write the TSOā¦
Commercial air is safe for a long time, no-one is getting on an airliner with no human back-up on board - the risk/reward service figures donāt make sense.
After a few years of car automation, and potentially something with enormous economic and social impact, will be automating trucks. 24 hour long-haul automated trucking, with the biggest threat being highway robbery. I guess thatās when we get robocop?
Itās not about safety primarily, itās about personnel costs.
I doubt that we will see autonomous cars before we see autonomous airliners. From a computational intelligence perspective, the car problem is immensely more complicated because cars operate in such cluttered environments. We are still a LONG way off from truly autonomous cars, despite what Tesla would like to make you think. They have constrained the problem of autonomous driving to the easy tasks, the things they have omitted are immensely more complex and will require quite a few years more of research.
On the other hand, the problems associated with flying an airliner from A to B (paved and well equipped airports of course) are mainly solved, as far as i can see. The biggest problem here would be handling emergencies, however given that state of the art airliners have been getting meticulous system sensor coverage, this is IMHO a problem that is solvable without the need for technological advancement.
I disagree, in that for a service industry (where people pay for tickets and governments regulate) then the rational answer of āit is about costsā doesnāt apply 100%. If you could buy a ticket for 50% cost where there was no humans capable of dealing with an emergency on an airplane would you buy one? Would you buy one for your kids? After a single accident where a human could have prevented it, what happens to the stock of that airline? Itās not the ādriving the airplaneā thatās hard, itās the market around it and the peopleās perceptions.
For car and truck automation, it doesnāt have to be 100% solved. Itās like the ālast mileā problem of deliveries, in that automated trucking will start out modelling economic paths like rail routes. Most large scale truck deliveryās are set up between factories/produces and hub warehouses already. Itās not the case of being able to drive down a country lane in the snow, itās more the I-5 between two junctions (say Canada and Mexico )
Yes, looking at Cessnaās failed Skycathcher and how the price rose during its short life is but one example of how expensive aircraft and related equipment certification has become. I went to buy some loctite red at an aviation parts retailer recently and a small bottle was something like $60.
its all about the money, if people start to smoke the same amount of cigarettes per day as they shoot bullets per day then nobody will care about the cigarette business and will just make it illegal.
people spend too much money everyday on cigarettes and that counts.
and back on topic, I agree with the fear as mentioned by @BeachAV8R
If my airplane crashes i really donāt care if thereās a human or a machine in control. Of course youād have to make the AI robust in that it can deal with system failures to the same or better degree than a human pilot can, but as i said in my previous post, i think this is possible with the current state of the art. For cars it is entirely impossible right now.
You still fly with airplanes after that Germanwings guy intentionally drove his plane into the French Alps, i presume. If it is at least as safe as a human pilot statistically, airlines wonāt bother with human pilots. No more pilot strikes, just imagine the stupendous sum Lufthansa could have saved in 2016. You really think thereās an argument to be had?
Yeah but still, obstacle classification is damn hard. Remember the Tesla that thought the truck it rammed into was a streetsign? Or the accounts of Teslas switching into the opposing lane? Turning your airplane argument around, itās not about driving down an empty perfectly straight road. The rate of āsomething unexpectedā happening in road transport is way higher than in airplanes.
Interesting tangent. Iād say itās nobodies business whether somebody wants to impede his/her health by smoking. As long as tabacco manufacturers (and therefore indirectly smokers) have to pay for the additional costs that their habit imposes onto society, itās their own responsibility. Oh and of course non-smokers should be protected from having to endure passive smoking as much as is reasonably possible.
Thereās zero chance of you being able to take a commercial flight in the next 5 years without a human pilot on board. Thereās a pretty good chance you can get into a car (probably a cab) with an autodrive ability and no human driver in that time-frame. Itās already happening.
When a truck crashes it barely makes the local paper, when a commercial airliner with 300 passengers goes down itās a major event. More people die on the roads, but an aircraft disaster is big news. Itās because when an airliner catastrophically fails (including AI or human) the consequences of that are bigger than if a truck crashes. Being in the air is different from being on a road.
The tech isnāt the issue, itās because of peopleās perceptions and that canāt be hand-waved away. The commercial reality is that automated driving, especially for delivery and long-haul trucks is a huge economic disruptor, while having no airline pilots is technically possible but aināt happening soon.
I donāt think itās something any of us have to worry about, even myself being fairly early in my career.
With how slowly the government procures assets and how much they have invested in new F-18s and F-35s, Iām not worried about being replaced anytime soon. I could, however, see use of something similar to a predator as a compliment to the airwing happening fairly soon but Iād guess thatās still 10+ years away for a carrier operated combat UAV.
For a lot of the reasons above, I donāt think commercial travel will be effected for many years either.
I agree that you wonāt see an airliner in that time, that much is a given with the development times of airliners. But from my experience in the car industry, autonomous driving is very unlikely to happen in the next 5 years, especially not in populated areas.
Not commercially. There are extensive tests, but commercial application, especially in densely populated areas such as the typical cab ride scenario are ways off. The stuff that the Teslaās feature right now is an assist. Itās nowhere even near the capability of autonomous operation.
I might be a bit special in this regard but i read about each Google car crash shortly after it happened.
Hmm, iām not sure what you mean. If the tech isnāt the issue, why is there so many computer vision departments busting their behinds about this?
5-10 years is typically the gap between cutting edge research and industry application. The computer vision guys i know certainly do not regard the issues associated with autonomous vehicle operation as solved so thereās hardly a chance youāll see large scale application in that timeframe.
The tech isnāt the issue with regards to why pilots are needed on airliners.
The computer vision guys you talk to are betting against Google, Tesla, Uber and a few others with deep pockets - perhaps theyāre annoyed they didnāt make it past the first phone screen?
Anyway, automated driving will take a while, but like I said a few messages back, itās not an all or nothing, itāll be gradual - unlike something like 100% safe completely automated commercial air travel.
Well let me tell you, Google, Tesla, they only cook with water. What they have accomplished is no less than impressive, but they are hype machines too (especially Elon Musk).
In case you havenāt noticed, Google has spun off its autonomous car project (they are now called Waymo i believe). They will supposedly try to enter into joint ventures with established car manufacturers. Weāll have to see where it goes, but they donāt have a finished product and will have to turn a profit soon, because that Google money tap is probably closing.
From what i gather, Uber is about at the same point that Tesla is at right now (assisted driving with the human driver having to keep his hands at the wheel at all times). Thatās a fair way off from real autonomy.
Oh and if you still think that the industry is a technology leader in this regard, those Uber Engineers were hired off of CMU
You could be right, itās no fun being the one disrupted - a lot of old industries have historically not enjoyed the process.
The Uber Pittsburgh trial is automated, but with an engineer on board to watch at the moment. Iāve personally been in a Tesla driving itself on a highway and the feeling is quite surreal if youāve never done it.
Anyway, back on to @chipwichās topic about pilotās jobs.
Pursuent to our collective 2017 health initiative, I worked out this afternoon - which gave me time to think about this a little more deeply.
I suppose my one of my biggest fears would be that in 8 - 10 years, when my oldest begins thinking about flying leasons, that I wouldnāt be able to rent a two seat trainer for introduction or polish, because general aviation as we know it, doesnāt exist. But when one considers the age of your typical training fleet, how many FBOs are run at slim margins, that we have military aircraft older than their flight crews, not to mention all of the small operators, and the populatiry of experimental/warbird/vintage aircraft, I probably have a much bigger chance of not passing a flight physical than unmanned aircraft systems proliferation.
Iām not a real aviator, but I did try to give this some thoughts.
Just how much of a cost factor are the pilots during a flight? I know good pilots can earn in the five figure range, and you need two of them. Factoring in other variables like retirement, a place to stay overnight and healthcare having pilots on the roster is an expensive ordeal. But I saw a documentary about an A320 (oversized example, I realizeā¦) that had to return just after take-off and dumped $19 000 worth fuel to meet the landing weight. Can we really expect HUGE savings by factoring out pilots?
Wouldnāt advancements in other sectors reduce costs more than fully automating the pilots job? Iām not saying developments in these sectors would halt just because of that. But fossil fuels are only going to be more expensive and there doesnāt seem to be any viable alternative for aviation any time soon. The Concorde had an ingenious yet laborous system that allowed it to pump fuel through itās fuselage. This allowed the Concorde shift its CoG and therefor, trim pitch without relying on draggy trim tabs.
and who in their right minds would step on a plane with only an automated pilot. Iāve seen to much air crash investigation where the comms were one of the first things to go down. I guess for a plane like this it would mean that it would go either ballistic, or worse, rely on some sort of windows help to try and figure out the problem
I think the cost of the crew is a drop in the bucket. When I was flying the Challenger 604, I calculated once that it would take 50 years worth of my gross salary to buy the airplane at its current value (for a 10 year old airplane). That does not include anything else that goes with operating an airplane such as fuel, hangar, maintenance, insurance etc etc. Just to have it sitting on the ramp. For the airplane I currently fly, that figure is over 200 years. I guess I had better start saving .
Regarding automation, there are some things that a computer will do better than a human, and some situations where the opposite is true. Most aircraft accidents these days can be attributed in some way to human factors. Mechanically aircraft are extremely safe when properly maintained. A pilot having a bad day is a definite weak link, but then you get the situations where a pilot thinking outside the box has saved the day where I donāt believe that a computer would have found a solution.
)
