Aug. 13, 2018, © Leeham News: The bizarre theft of a Horizon Airlines Bombardier Q400 at Sea-Tac Airport Friday night by a 29-year old employee will take some time for investigators to unravel.
The employee, a ramp agent, appeared to have no other motive in mind other than a last joy ride before ending his life.
Questions about how he was able to have solo access to the airplane, which was parked in a far remote cargo area of the airport, start it up, taxi without challenge until turning onto a runway, taking off (apparently without fully releasing the brakes) and perform aerobatics before taking his final dive into the ground are just some of what investigators will look at.
Right away, news media asked how could this security breach happen?
The answer to this one is pretty obvious. When an “insider” decides to do something, it’s virtually impossible to defend against it.
Law enforcement officials have been living in fear for decades of the home-grown terrorist. We here in the USA have had success in deterring foreign terrorism on our soil since 9/11. There have been many incidents of home-grown terrorism that proved impossible to detect in advance.
In commercial aviation, there have been a few examples of insider threats going undetected.
Two high-profile incidents come to mind.
In December 1987, an employee of PSA, an airline that had been purchased by US Air Group, but which not yet had been absorbed into the US Air brand, was fired. His identity badge was not immediately confiscated.
This ground employee used his badge to bypass security. He boarded a PSA British Aerospace BAE 146. After the plane was at cruising altitude, he shot the supervisor who was a passenger. He entered the cockpit, murdered the pilots and the plane crashed, killing all aboard.
In April 1994, a FedEx employee was dead-heading on a DC-10 flight. He was facing dismissal and decided to hijack the airplane and crash it. It took all three cockpit crew members to fight off his attack, subdue him and—nearly incapacitated themselves—land the airplane.
The Horizon incident really demonstrates why the idea of allowing airliners to be flown by a one-person crew is just nuts.
There’s a provision inserted into the current US Federal Aviation Reauthorization bill to study allowing solo pilots on cargo aircraft.
It’s a mystery who’s behind this. Nobody owns up to its inclusion.
It’s an incredibly dumb idea, pilot shortage notwithstanding.
There have been several incidents (some disputed, some not) of investigators concluding pilots deliberately crashed an airplane, taking everyone with them.
Germanwings Flight 9525 in March 2015 is an undisputed example of a solo pilot in the cockpit deliberately taking control of the airplane and crashing it to commit suicide, taking with him all aboard. The captain had left the cockpit to use the bathroom. The co-pilot refused to open the door for his return.
This is another example of the insider threat, though it turned out there was plenty of advance evidence the co-pilot was mentally disturbed and never should have been flying. This is also why when one pilot leaves the cockpit, a flight attendant steps in—so the pilot is never alone.
Egyptair Flight 990 in October 1999 plunged into the Atlantic ocean. The flight originated in the US, so America’s NTSB was the lead investigatory agency. The conclusion: pilot suicide, something Egyptian and airline officials dispute to this day.
Malaysia Airlines Flight 370, the one that disappeared into the Indian Ocean, is suspected by many of being a pilot criminal act. Others think it was a hijacking. Sleuth Christine Negroni believes pilot hypoxia is responsible.
Then there’s the much more common need for more than one pilot to handle an emergency.
You will never convince me one pilot and computer back-up—or some remote pilot on the ground—could have helped land US Airways 1549, United 811 or United 232. The problems were too numerous. The workload was immense. Just ask the five pilots on Qantas 32, the Airbus A380 (a two-man cockpit) that had the Rolls-Royce engine blow up.
The first one, 1549, had a 100% survival rate.
The second, 811, lost nine passengers who were sucked out of the airplane in-flight when a cargo door failed, ripping a huge hole in the fuselage of the Boeing 747. With two engines out and overweight, the flight crew coordinated to successfully land the crippled airliner with a 100% survival rate of the remaining passengers.
United 232, the emergency landing at Sioux City (IA) with famous footage of the crash landing, required coordination not only of the three cockpit crew members. A fourth pilot flying dead-head came to the cockpit to help. The crash landing killed 111—but 185 survived, thanks to crew coordination.
Qantas 32 made a successful landing, with a 100% survival.
Finally, there’s flight deck crew incapacitation.
It’s rare but it does happen when one of the pilots becomes ill, has a heart attack or dies in flight. The remaining pilot lands the airplane. In one instance, the pilot asked a flight attendant to join in the cockpit to assist with some of the workload.
If a solo pilot becomes ill, incapacitated or dies, then what? Yes, a remote pilot can land an airplane—if he knows what’s what.
An automated on board system can land the airplane. Category III landings do this all the time. But how will it know?
The Horizon incident last week didn’t involve a pilot, but it certainly brings to mind just how stupid the solo pilot idea is.
I agree with you.
Bu the opposite could be a good thing, as examplified: germanwings 9525, MH370 or Egyptair 990 could have been saved by automated protection and/or remote control (or this horizon Q400 theft). Why not explore the possibilities? There was a debate on aviationweek, with opponents and proponents.
Lets look at “automated protections”
You have to program every remote possibility of any action, one flaw and the aircraft crashes.
When does it kick in? Ok, monitor the pilot fully. They can’t even tell if a patient is dead without extensive scans of the brain, so we put in an MRI in the cockpit?
Then the system goes haywire and it takes over from the pilot?
I do system programing. Its beyond incredible how complex they can make a simple function. And taking control is not simple.
You might do it with a Cray Super computer, taking up half the aircraft.
We just need trained and decently paid pilots.
People today think and app is acualy reality. Behind a functional app is a huge amount of real world work to make it do anything.
All the App is is the face of a background of major technical operations.
While I agree in general, I do believe that the economic pressure will come up with some other variations than a fixed dual cockpit.
I assume the (not being a pilot myself) workload during take off and landings are biggest. I could see that a first officer has “double” duty serving the passengers during the flight before heading back to the cockpit for landing.
And purely from an economic perspective, having always two flight crew for those 0.x% of flights where you REALLY need more than one crew, is kind of a “waste”.
And if a single cockpit becomes reality, then why bother with other redundant systems on board? That could make aircraft much simpler, lighter and cheaper.
But don’t get me wrong. Personally I wouldn’t board a commercial airliner with single pilot. I’d happily pay a premium for that extra bit of redundancy.
Such is common with flights in the Canadian North: The First Officer is expected to give a hand with boarding passengers and checking baggage but ‘switches’ role completely when boarding is complete. But this of course is only valid on aircraft such as a Twin Otter (in passing my most delightful vision of pilots working together is when – in a Twin – one can see them cross hands to ‘duopoly’ the settings of the flaps on the cockpit ceiling controls!)
Silkair 185 is another crash where the NTSB found the cause to be pilot suicide (Dec 1997).
I think the one pilot studies build on full automated control meeting current requirements + a pilot as a kind of back-up. Plus a 2 / 3rd pilot being on the ground if required.
Space flight is fully automated. People are much too unstable, unpredictable and unaccurate to leave to mankind.
If the aircraft had some build in (enveloppe) protection against crashing, or someone one the ground had a “return to horizontal flight button) maybe on e.g. Egyptair Flight 990, Germanwings Flight 9525 would have bee different.
On Airways 1549, FBW was keeping the aircraft stable / nose up/ flat during the ditch, updating everything 30 times a second, with all available controls. Nothing bad on the accomplishment of Sullenberger, but probably a human couldn’t have pulled it off.
No place for romance, emotions. Authorities wil do objective research & safety is #1.
Astranauts are not unstalbe.
In this case, you have a window that has been pre cleared, the Rocket does not go until they push the button.
Its a absolute specific route to space that tolerates no deviations and a person can’t do that at 3000+ mph.
But its all absolute known, tarty to have a rocket that self launches deciding weather, aircraft overhead and dodging them as it goes up and ?????????????????????
As long as the reaction time is do-able, people are the best mobile computing device anyone has come up with.
Space flight isn’t even close to being fully automated. The Apollo moon shots and the Shuttle were manually flown for the most part for example.
Shuttle entry was not manual, it was automatic and shifted to manual.
Time counts for burns are done by computer.
But while it is a complex calumniation, its a pre calculated one and then to keep on a trajectory.
Satellite launches are totally automatic.
Shuttle going up is automatic.
A human being cannot begin to react fast enough.
But its also very single goal driven, flying a jet aircraft in the atmosphere gets into routes, timing, other aircraft, ATC, landing.
That is aside from an other take over control idea.
Shuttle OMS burns were manual. Docking with the ISS was done manually as with rendezvouses with the Hubble. While deorbit and landing was run automatically until approach reentry could be flown entirely manually (tricky in a couple of areas though) and large parts were on STS-2.
Most cars today come with key Fobs – why don’t the airframers include this common technology for aircraft? If not that, authorized Face-id, fingerprint or retinal scan locks could be implemented.
Ah, STOLEN key fobs?
That is why I offered the Face-id, retinal scan, finger print option. If I can keep my kid off the iPad from playing excessively with the flight simulator – we can keep unauthorized individuals from operating an aircraft 🙂
Its called a single point of failure.
If the contact does not work (relays or the remote) no big deal.
Any loss of fob and your airplane quite? Not at all good.
Sit down and figure out a way for a non single point of failure (actually N+3) and then come back with the answer.
Thanks for a very thought provoking article, Scott. However, I think overall you’re “swimming against the tide”. Technologically, as you know, we’re seeing very rapid advances in artificial intelligence and vehicle automation. See Lt. Colonel Jason Hansberger’s August 8, 2018 opinion piece in AW&ST. (He’s the 89th Air Wing’s Chief of Safety and a C 32 pilot. Impressive aviation credentials.) As he notes, it’ll probably start with military and freighter aircraft. Once “proven” in these lines of aviation, it’s on to LCC and ULCC carriers. As Billy Joel sings, “It’s Just A Matter Of Time”.
After the Germanwings Flight 9525 in March 2015, the talk was maybe planes in emergency situations should be taken over by a computer, either in the plane or on land. With more ATC being handled by GPS, it seem like the technology is close.
And who decides?
If it can be actuated by remote it can be taken over by remote.
Write up a fool proof system line by line.
That was the talk at the time. The co-pilot had major things going on, and the company knew it. But say your at 380, and it is now a GPS system. Suddenly, the planes at 350. Bingo, the computer catches it. Either the onboard safety computer contacts the flight control or the remote contacts the plane. Now they’re at 300. No radio communication is established. At this point, an emergency assessment is made. 250. Action. The cockpit no longer has the con. The plane is flown by sources outside of normal control. Something like that could have save some of these situations we’ve seen in recent years.
“Dave, I have the greatest confidence I can complete this flight.As you know the HAL 9000 series has a perfect safety record…”
9 seats aircraft can already be flown with single pilot. The risk would indeed be higher but what would prevent the same to be applied onto 19 seats aircraft? What’s the difference between a beech 1900 carrying 9 seats and a beech 1900 carrying 19 seats?
people make errors..
Yes they doo
And people can adjust to errors made.
Granted some don’t, then we have AF447.
Who makes the equipment and write the program.
It is a bit easier to design a preprogrammed flying system on a fly-by-wire Aircraft by only letting the pilot taxi to line up or taxi after landing and on the taxiway.
The flight is automatic connected to a Boeing/Airbus computer network hooked up to ATC and Airport Towers.
The military will be first I assume and make the system safe to a certian level. When it is safer to fly with no pilot on board than take a cab to/from the Airport the switchover will happen eventually.
Then attacks can be done remotely, by an insider, after some security lapse.
We need something inside the aircraft that can deal with anything short of total control lost. Right now it’s human pilots, but in future, we may have AI pilot. As with self-driving cars, we need to either all-in (full AI pilot 100% time) or all-out: human pilots in full control.
I think hacking into it gets harder and harder. Try take control of a Global Hawk in flight. Never been done as a hacker attack as I know of. You need to down 15-20 serves that have both satellite and ground connections to the aircraft to shut down communication. The aircraft only need a few millisec of communication now and then to make selections of its preprogrammed routes that is not done with ADB-S in/out. All of us will see what Boeing will come up with at MIT in a few years time.
My point is that even with remote control, if the control is by human, the problem with insider is still there. We need full AI pilot that installed on-board, not remote control by external system.
If airplane can be controlled remotely, attacks can be done remotely too. Or by a relatively simpler Denial-of-Service attack that overpower the control server.
As with self-driving car, I think we need real self-piloting airplanes: airplanes that have pilot onboard that can deal with any problems can arise. Any remote-controlled systems will have weakness that can be exploited.
Oops, somehow my answers gone missing. That why there’re so many of them!
I am having the same issue.
I can’t stay logged in as the option is no longer there and any log in is on a subscription level.
I have to comment, go back latter and see its come up.
Edit function does not work if it does not come back up soon endogenous ergo more errors now.
Always read your interesting blog without commenting. Now as a mathematician I have to protest however against your logic.
First, you use examples of the present two pilot situation to argue against an one pilot situation. There is a logic flaw here. Why would there not be other options that are as safe (as some other reactions show).
Second, looking at the examples you give a more logical line of reasoning would be that having one pilot would half the chance of pilot suicide. This supposes that when there are two pilots the suicidal one will always find a way to put the sane pilot out of action. This hypothesis seems supported by the facts since we do no know of pilots that have prevented a suicide (though you may argue that the mere fact that there is a co-pilot already has a preventive effect).
To the best of my knowledge, the suidices have been at least started by a single pilot while the other one (all cases) was out of the cockpit.
Egyptian is the exception where the pilot got back in but should have clobbered the co pilot with an extinguisher (clearly he did not believe what was acualy going on, hard to adjust to) – I believe that was before locking cockpit doors
US requires a second person in the cockpit at all times but does not have to be the pilot.
There may be some credence to the second person in the cockpit as a psychological deterrence.
Clearly not alwyas physical but the physical attacks were external.
Ironic if you count up the dead from the air crash itself, we are likely due to more dead due to locking cockpit doors than the terrorists have inflicted. Bombs don’t count of course.
My view on a lot of this stuff is 100 years from now, assuming that our civilization survives, our descendants will look back on this completely differently.
We think about critical tasks, like piloting, or surgery, or whatever, and we instinctively think “that’s too critical to be left to a machine”.
100 years from now, they’ll look back and say, about the same tasks, “that’s too critical to leave to a human”.
I also think we’re not there yet. In theory there could have been a system that would have allowed the authorities to take control of the Q400 so that it couldn’t have taken off, or if it had taken off, forced it to land.
But then the issue is what happens if the bad guys get control of that system. The Russians apparently were all over the US election of 2016, and none of those systems have yet been secured. In fact, you could make an argument that US elections, right now, would be more secure by going back to paper ballots. So until we have relatively foolproof ways of securing systems (which we don’t), putting an aircraft under some kind of remote control creates as many problems as it solves.
But that doesn’t mean that will always be the case.
enpalned: The Russians did not take over the US election system.
They used what we used to call 5th column in the form of Twits, Space Book, etc.
No vote was changed by hacking.
How many votes were influence by the propaganda is open to debate but the Agent Orange wins in key states was under 100,000 total.
100 years from now we may have the low cost sensors and systems to have auto flight.
But so far the driver less cars have not been a great success.
If you can get a jet fighter onto the approach to an Aircraft Carrieer, auto system do better than pilots.
But its an extremely focused and limited task.
No one has put out a fighter that can attack and defend itself.
Most experts don’t want to let an aircraft try that.
There is a middle way regarding fighter jets, that a leader flies the manned jets but the rest of the fighter group is unmanned and follow his commands. It is a bit like fighetr jets worked in the 40’s-50’s where the leader took them to the action, assign tasks and told them were to meet up and guide the survivors of the group home.
Now it can be similar tactics where the Communication is digital and the rest pretty similar besides the unmanned fighters can pull +25G/-15G’s and have big sensors putting their sensor inputs into the big battle management system that then allocate targets to the Group leader that assigns them to his favoirte unmanned wingmen. Think Dassault is on to something similar.
Scott: I had not hear of the 1987 incident.
As for MH370 Hypoxia, its amazing how many course changes of radical nature you can make when your brains is frozen.
Yes how many precise course changes! I honestly don’t see how in the world anyone can make the hypoxia argument.
There is another guy who feels it was a raging fire in the cabin, the Co Pilot heroically crawled form the back to up front and made all those control changes (with portable oxygen and no arctic clothing) – oh and he was flying ti so they would not crash and kill someone, right up the Straight of Malacca, past Panan and a number of other large cities.
Can you say right turn, dive? Nope, over land, by cities and then down the outer coast of Indonesia and on to never never land
And then the aircraft did not bobble the least bit.
It is amazing what people can convince themselves of.
Clearly one of the pilots is a parallel large scale killer.
I have to laugh as I do programing on equipment.
Its like programing a house, there is alwyas the exception to the rule.
It turns out the best way to deal with it is for a person to turn the lights on when they use a room and turn them off when they leave.
Its not that it can’t be done, but the variables are worse than a Chess game (which is played in one dimension sans 3d chess)
It took them 50 years to do a program that can finally beat a good chess player – and that i on a two dimensional board with extremely limited ops for the pieces.
Oddly, there is nothing more capable than a human being to adjust to variables (humans also are fully capable of getting tunnel vision per AF 447).
Aint nothing perfect.
In that tragix one, a skilled, experienced 3 person crew should have simply fainted & the aircraft would have done the right thing..
Yep, in the FA-18 pilots are taught when they lose control to simply take their hands, grab the windshield and let the plane fly itself out of trouble.
Thats an intering one. A fighter that makes the best of it if the crew deems the risks unacceptable & bails out. E.g. trying a landing anyway,
making a belly landing in empty terrain, avoiding populated areas, crashing into the main target.. wathever makes most sense.
Also to be noted that most of the incident take place at low altitude and its of no use.
Also the system requires pilot input.
They are working on an automated recovery system tested in the F-16.
That is a very limited specific task triggered by a data base of terrain and the aircraft position.
But fighter pilots are well trained in action if the system goes off the rails.
What happens for a civic system if it goes zark?
external systems are subject to hack and other failures.
Internal has trigger issues.
Very large bandwidth needed.
You do understand that there are thousands of single pilot airplanes (including business jets) throughout the world today and single pilot operation (SPO) is not an issue.
Instead of rejecting SPO outright, maybe we should define what we would need to see as future technology to improves the odds of safe return.
The discourse above seems a little like the discussion that happened when we first removed the navigator from the cockpit and provided enough info to the pilot to navigate… The same happened when we removed the Flight Engineer.
We are on our way to ‘no pilot’, let’s define a safe way to get there.
A single pilot airplane with 9-10 people max capacity is different than an airliner. My comments stand.
Lets wait awhile and mature the technology.
Its not happening in the next 6 weeks.
I think you will find SPO operations are a lot higher incident/crash rate
Ratings and experience are not the same let alone the operations.
There is a reason regionals get pilots for less.
Then you have local are flight ops that do not fall under the larger Commercial (have yet to see a C206 flight simulator)
To have a local air company operate to the standards of bigger commercial makes it cost prohibitive (pretty rough as it is)
Of course we could just make it common and pay the price (ie put the whole SPO ops out of business)
And that does not include non commercial SPO.
Comparing Apples to Pomegranates
For every experienced ‘Sully’ who saves a plane load of passengers there are experienced ‘noddies’ who lose one when they shouldnt.
The accident stats are full of pilot errors and its not just beginners
They will have to invert the way SPO works. Instead of it being starter jobs, those should be with dual pilot operations. The SPO should all have say a minimum 5 years or more experience or equivalent in flying hours
This will only help to further increase the stupid security checks and pre-cautions which keep piling up since 9/11. A pure bullying of 99,9% aviators and still never strong enough to keep the evil from doing horrible things. Yeah, go ahead, make things more complicated with unreliable face ID, buggy Software Protections, and “oh I forgot my coded RFID key”. It will happen again No matter how complicated you will make flying to all other People. Let’s finally grow up and face it.
Nah, we can’t have that.
We can make it 199.9999999% safe.
And then we kill tens of thousandth a year in hospital screw ups not to mention car crashes.
The first question I had after hearing of this incident is, how do you stop a rogue aircraft on the ground from taking off?
I’ve seen a number of such scenarios played out in Hollywood content, but is there anything that can be practical?
popup barriers at runway entry points under tower control would have done the job to keep him off the runway and would probably be the cheapest 90%+ solution
now do the math, ~300 largest airports in the US airports at ~20 runway entry points per airport $200-500K per entry point, ~$1.2-3 Billion to roll that out across the country == never gonna happen
You can always block the runway with a vehicle but it takes alert security.
Yep, guard for a once in 10 billions flights thing.
Said guard goes to sleep and it still happens.
Life is not an App, there is alwyas uncertainties and possible bad things.
Well the guard could be guarding against more than one thing. I mean it would be nice if they had regular patrols for general security period.
It all costs – there is a balance. Do you spend billions on a once in 100 years incident?
And as one guy noted, would you want to drive in front of two spinning props?
I wouldn’t worry about being a passenger on a single pilot plane. In an emergency there isn’t a plane in the world that I couldn’t put on the ground! 😀
Agreed, it would not be pretty but I believe if I had assist, landing an A380 would be doable.
I got to fly an MD-11 simulator, did a nice landing with no time on the controls. Basics still apply.
Sim Instructor was quite delighted as the previous flyer had stall warning going all the time.
Clearly a lot of the controls and setup would be beyond me.
Any small aircraft I alwyas sit in the right front, if its people I know they insist on it.
Made some flights around Hawaii, pilot was good, also quite old. Everyone was happier to have me up front with him.
Well I wasn’t really talking about “landing” per se, just putting it on the ground!
Controll inputs from the ground?
Imagine 9/11 with no high jackers on the aircraft just some guys sitting somewhere in front of a computer with about 20 aircraft under their remote control.
Several airlines change pilot and copilot on a regular base so they don’t get used to each other and develop “bad” habits. How does a computer react to bad moods of a pilot? Does a computer even recognize it?
“Dave, stop. Stop, will you? Stop, Dave. Will you stop, Dave?”
Finally I think it is a mayor psychological support that there is sitting someone right beside you working with you to solve the same problem: to walk away after landing.
It depends on the crew.
Clearly some are so dysfunctional that it does not work at all (AF447, Air Asia A320, Asiana 214)
The biggest issue is lack of the right training and pilots understanding the fundamental.
AF447 should have know that if you are still alive, you have not gone to zero speed at 35k, Even Everest is lower than that.
An American 757 pilot did just the opposite, pushed down when he lost speed (what you8 should do if its really lost speed). But he was 20k and nothing around and if you come to a sudden stop, you are dead.
They are starting to work on training that finds out if a pilot really understands that or just panic reacts.
Reality is that there are a lot of pilots who should not be.
There are likely people who should never be pilots (its just a skill) as they can’t overcome that panic issue and just react.
People are weird beings. Little Green Men and MH370 Hypoxia (of which littel green men are more likely)
This is a simple case that will be Verified as such when the dust settles. This employee was authourized to be with planes alone. He was ill with depression that wasn’t dealt with or was , but not properly. What he did is a desperate act of a human who had to be suffering beyond what most could comprehend.
Horizon nor any other airline can do anything legitimate to protect against the theft of a plane from a remote apron by a security authourized person with the determination to do it. I hope everyone calms down and doesn’t start implementing changes. No changes to security or operations will mitigate or prevent this from happening again under similar circumstances. This was an isolated random tradgedy that fortunately involved the loss of one life and a plane. No security breach, no terrorism, just a guy that chose a dramatic way to exit life and end his suffering.
Diagnosis of depression seems a bit premature.
Certainly off what is deemed to be normal.
But then at least one person believes you can pilot an aircraft with an Oxygen starved brain for over 7 hours and doing some smooth flying.
Makes for an interesting question as to how sane human kind really is.
Makes you wonder why pilots are more likely to think they are so important that they can take a 100+ people down with them?
For comparison, how many times do you hear of bus drivers or train drivers taking their own lives by crashing and killing passengers.
There are many times more bus drivers (times 100?) so this should be fairly common if parity in the psycological factor would exist.
I don’t mean to say driving a bus is as complicated as flying a plane.
Or another example, how many times does it happen that a soft drink company employee that’s about to get fired poisons an entire batch of coke?
Interesting questions. Why do we let 10s of thousand die each year in hospital screw ups? Or car wrecks.
I suspect its the wow factor.
Same with mass shootings.
Back to the basic question, is there a tendency in human being to be only functionally nuts and a certain number trigger when the opportunity to make a big splash occurs?
I know religion is taboo, but the overall concept and belief system requires some serious mental juggling. Its not a belief as such, its trying to make it logical. That has got to strain a lot of cells.
Maybe with all of societies contradictions (rule of law when in fact there is none) finally something snaps?
A democracy where electoral collage based on a weird 1700s idea of balance decides that a minority candidate wins?
Let’s drop political, mass shootings, religious conversation. This is about aviation.
I will comply of course
In this case its trying to remove a Human Aviation issue from its context, those other non aviation aspects are part of who and what we are.
I really didn’t want to go in the direction of politics and or terrorism. And strictly interested in the psychology behind suicide by a pilot while flying (not the terrorism kind, though there are interesting psychological factors in that area too).
Understood, but any problem with pilots or in this case an aircraft theft of that type, gets you into the human aspects areas.
Unless you understand it and also can detect it, this sort of thing will keep happening (though a stolen aircraft of this type by a non pilot is a one off)
When DARPA ran the first Grand Challenge, not a single vehicles came close to finishing the course. “Experts” said the course would never be completed. It was completed. Same with DARPA’s Urban Challenge.
Waymo has had over 400 people, aged 9 to 69, using fully autonomous taxis (without a safety driver) for over a year in Arizona. They will start the rollout of 60,000 robo taxis this year.
With AI, machine learning, etc. we are approaching the bend when the line goes near vertical (5-10 years) and it will be a shock to most people how quickly everything will change.
We also have incidents of autonomous cars not doing what they are supposed to.
And like Chess, its a two dimensional world that you would think would be quite easy to have sensors for and programing.
Me thinks we need a whole lot more data, like about 50 years.
Robo-vehicles already have a lower “incident” rate than human drivers, and the robo-vehicles continue to improve while human drivers don’t.
The two dimesnional world of robo-cars with the ever changing dynamix mix of vehicles, pedestrians, unleashed pets, playgrounds, multi-lane roads, complex intersections, road regulations, etc., etc., is a far more difficult environment than airliners flying autopilot in protected controlled airpace.
Waymo already has over 8 million miles of self-driving experience – that would take an average american driver over 300 years to complete. Waymo is now logging over 1 million miles a month of robo driving. And this is on top of the 2.7 billion miles driven in simulation in just 2017. In addition, they are using an abandoned air base to throw the most difficult, improbable edge cases at a test fleet. And this is just Waymo. All the others are doing similar.
Well as a in the real world technician, what I can tell you is all those ideas in a test environment sound fine, then they put them into operation and not so much.
As has been noted with small aircraft, a mistake and 9 people die.
with a car, a mistake and it may be a dented fender, injured and a really major one, 4 people may die.
A mistake with an aircraft (large commercial) and 150+ die.
Not to mention the failures in ATC when it becomes a non controlled environment let alone one on autonomous by one set of rule and a human being with the other.
Like open rotor, its 10 years away, continuously.
Air Failures are big publicity events, its going to be 50 years before you see single pilot let alone autonomous.
Just talking Waymo, over 8 million real world miles and over a million miles added each month. And soon multiples of millions of miles added monthly with the rollout of their 60,000 robo-cars starting this year.
Boeing, Airbus and Embraer are already working on developing airliners flown by AI. By 2030, there is no reason why pilotless RJs wouldn’t start to be introduced into service.
Having looked at Waymo and who is behind it (as well as Uber or Facebook) not what I would call models of good citizenship let alone admit any mistakes.
Just something out of the blue here, what about a “land at nearest airport” command. A pre-programmed command that includes a list of acceptable airports, determined by the airline, where the aircraft can do an auto landing. If hacked, the worst that anybody can do is cause an unauthorized, unplanned (and expensive) landing but there should be no risk to passengers and crew.
Your comment isn’t so out of the blue… there were several systems developed that address your post. The “Boeing Honeywell Uninterruptible Autopilot” was one of them. Rockwell Collins also developed a system. The Honeywell patent was granted in 2006… 12 years ago. So we obviously have come much, much further in both software (AI & machine learning) and hardware. Internatioal Data Corporation calculated that corporate spending on AI and machine learning will go from $12 billion last year to $56 billion in 2021.
Boeing patents ideas all the time but does not build them (as do others)
I have yet to see a reliable autonomous car work in two dimensions let alone an aircraft in 3 dimensions.
For the past year, 400 families in Arizona have used autonomous cars in two dimensions and by the end of the year it will be available for everyone in Phoenix to request a driverless ride.
And the issues are low speed and two dimensions, hot 150 to 600 mph in 3 dimensions.
And its still a tech company that will lie through its teeth.
You might look at the recent reveal that they track your locaiton even if you turn it off.
You can be as enamored with Wymo as you want, I sure am not.
Comparing a slow speed good weather city in a car with an aircraft indicates a great deal of lack of knowledge.
I think there are less unstable pilots out there than “radical’s” that can hack a pilot less aircraft an bring it down.
In the industry I work in monitoring of pulse rate, eye blinking, etc. serves as early warning systems that it is sending on a continuous basis to an operations room.
The best of commons sense.
Cars, trucks bus…
– Highways = 100’s of vehicles up to 6 lanes wide – tailgating, changing lanes, merging, exiting, etc. etc.
– Cities = uncontrolled intersection, blind corners, distracted drivers, schools, parks, etc., etc.
Less complex to have autonomous airliners since airliners already can fly from takeoff to landing completely autonomously.
Plus airliners are limited to 250kt in Class B & 200kt in Class C and are protected by 15 minute longitudinal bubble, ATC surveillance and TCAS 2.
And it is not just Waymo… every car manufacturer and car sharing service is doing autonomous driving, plus many universities around the world.
As a professional programmer (fortunately in a less critical industry) I can assure you that you should never computerize something that can be none more simply. How about a simple key operated ignition switch in the starter circuit of the main engines? Without the key you can start the APU and tow the plane but the main engines are immobilized. Turn on the key and away you go. After the flight the pilots take the key to flight operations to give to the next crew. The automotive industry has decades if experience in the technology.