Duopolistic battle between two man-machine visions

Sometimes when I talk about driverless cars I am asked to what extent we already have driverless planes today.  The answer is a bit complicated:

The broader issue…is raised in an FAA report: “For any given situation, who will have final control authority?”  The pilot or the flight management computer?  Aircraft manufacturers and their automation designers have somewhat different philosophies.  Airbus has tended to favor the machine — its automation is designed essentially to prevent the plane from getting outside its safe “flight envelope” no matter what the pilot does.  Meanwhile, Boeing tends to give the pilot the final word — and its adherents can be adamant.  A Boeing-flying Delta captain puts it this way: “When shit hits the fan, a pilot should be able to disengage all the magic and fly the airplane with basics…All you can do is hope the software engineers haven’t screwed you with some magical sub-mode that, [sitting] in an office with a nice warm cup of coffee, makes sense at the time.”  For every fan of Airbus’s “make-it-impossible-to-crash” approach, there’s a proponent of Boeing’s support for new cockpit technology only where “there is no adverse effect to the human-machine interface.”

That is from Mark Gerchick’s Full Upright and Locked Position: Not-so-Comfortable Truths about Air Travel Today, a pretty good book although much of the material may be already known to some of the potential readers.

Here is a recent FT article about the advent of self-driving ships.


'The pilot or the flight management computer? '

Both. The computer systems can only operate within their designed parameters. Leading to something like this 1993 incident with two fatalities - http://en.wikipedia.org/wiki/Lufthansa_Flight_2904 -

'To ensure that the thrust-reverse system and the spoilers are only activated in a landing situation, the software has to be sure the airplane is on the ground even if the systems are selected mid-air. The spoilers are only activated if at least one of the following two conditions is true:

* there must be weight of at least 6.3 tons on each main landing gear strut
* the wheels of the plane must be turning faster than 72 knots (133 km/h)

The thrust reversers are only activated if the first condition is true. There is no way for the pilots to override the software decision and activate either system manually.

In the case of the Warsaw accident neither of the first two conditions was fulfilled, so the most effective braking system was not activated. Point one was not fulfilled, because the plane landed inclined (to counteract the anticipated crosswind). Thus the pressure of 12 tons on both landing gears combined required to trigger the sensor was not reached. Point two was not fulfilled either due to a hydroplaning effect on the wet runway.

Only when the left landing gear touched the runway did the automatic aircraft systems allow the ground spoilers and engine thrust reversers to operate. Due to the braking distances in the heavy rain the aircraft could not stop before the end of the runway. The computer did not actually know the aircraft had landed until it was already 125 meters beyond the half way point of runway 11.

As a result of the accident, Airbus Industrie changed the required compression value from 6.3 tons to just 2 tons per main landing gear.'

The software can be improved, obviously - but in this specific case, the software parameters actually prevented the pilots from braking the plane. Two decades ago, that is.

Boeing, however, remains decades behind in accumulating such experience. And it shows at this point - at least in terms of orders. http://en.wikipedia.org/wiki/Competition_between_Airbus_and_Boeing

For something more recent than two decades ago, supporters of the Boeing approach claim that the Airbus approach was responsible for the Air France Flight 447 crash, because the plane nose could be pointed upward and left there by the pilot setting the control stick and not having to subsequently touch it, as opposed to having to constantly pull back as in the Boeing approach, which is more similar to what one would do without fly by wire. Those who prefer Boeing's approach claim that with Boeing's method, the pilot and co-pilot would have instantly recognized that the problem with the flight was that the plane was tilted too far upwards (and thus lost lift) instead of realizing it far too late.

A lot of discussions about Boeing and Airbus control philosophies focuses on whether the pilot has "final authority", or sidesticks versus yokes. I think the issue you're referring to in this comment, the difference in speed stability between the Airbus and Boeing control laws, is much more fundamental.

The Asiana crash made me question the relevance of the Boeing / Airbus differences, though. Despite mechanically linked yokes, back driven throttles, auditory and visual warnings, and even the artificial feel system pushing on the stick with 80 lbs of force, the flight crew wrestled that plane into a crash. Maybe the moral is that both Airbus and Boeing are so safe now that most crashes will involve extreme outliers of poor crew performance, difficult to address from a design perspective....

This is true Matth. Who trusts the pilot, after the EgyptAir Flight 990 and this recent news? http://www.breakingnews.ie/world/mozambique-air-crash-intentional-617900.html

The other object warning that we get from Air France 447 is that computers' decisions are only as good as the information they get from their sensors. And if the pitot tube freezes over, the computers are flying blind and make correspondingly bad decisions. A second lesson is that the human pilots need to be trained and practiced to react correctly when things go wrong -- and the increased reliance on automated systems may be reducing pilots' competence.

If you'd like to see something recent from the industry on this issue, try David Learmount's post at http://www.flightglobal.com/blogs/learmount/2013/12/pilots-atcos-needs/ There's a link to a longer article that Learmount wrote for Flight International (access requires free registration)

The machinery on ships will need to be made much more robust and reliable, with much more redundancy before the crews can come off. There are so many things that can cause a ship to stop - fuel purifiers clogging or otherwise going down, aux boiler motor burning up, a couple cracked heads... It takes people to fix all these things, then there is the routine maintenance both on deck and in the engine room. Main engine work mainly needs to be done at the dock, but most of the generators, pumps, and other aux machinery is worked on underway. Deck maintenance is about impossible at the dock due to cargo operations so that would at a minimum require riding gangs from time to time.

Aside from the machinery issues there would also be significant navigation/collision avoidance issues. While there are very clear rules of the road meant to avoid collisions there are very few if any penalties for breaking them if there is not a collision. There are also vast numbers of recreational boaters and fishermen who do not know the rules and don't have to have a license to be on the water. Many accidents are avoided because of experience and the ability to anticipate what somebody else is going to do - this would be difficult to build into the logic of a computerized system. There would probably have to be many more restrictions of where recreational and smaller commercial vessels could operate, particularly in and around harbors.

I'm sure we will get there someday, but even if regulatory system was totally overhauled tomorrow I don't think ten years would be realistic for a real commercial application.

I'm not a FT subscriber, so maybe I repeat something from that story, but it strikes me that modern crews are already sufficiently small. Back in 2006 the (then) World’s largest Container Ship launched, with a crew of 19. How low do you need to go? At 9600 (20') containers, you are moving 500 containers per crew.

Most ships carry more than their required minimum safe manning. The cooks are not counted in the minimum, and there is often a larger than required engine crew because otherwise the work just doesn't get done. It is also common to carry "riding crews" - for example, a painting team may ride for a few months doing nothing but chipping and painting, or team will be brought on to overhaul the generators and aux engines underway.

One other thing I forgot about - reefer containers. These can and do go bad and the container ships carry a good supply of parts. If you lose a compressor or controller, or an outlet goes bad and you lose a 40' box full of plasma, high end ink, or some other valuable commodity you could easily have a claim that exceeds the crew wages for the voyage leg. This could also probably be largely overcome with a big investment in technology, if you look at crew cost the cost/benefit balance of the needed technological updates get fuzzy pretty quickly.

I could see the automation first in the dry bulk trades. Once loaded grain, coal and ore doesn't require much care or monitoring.

I'm ignorant of the details in this fight, so I'm agnostic. But it does strike me as backwards that there's a popular argument in favor of using humans as the fail-safe. I'm used to thinking of people as primarily doing activities that need to be backstopped by mechanical or computer safeguards, so the idea that humans (i.e. sleepy, lazy, intoxicated meatbags) are the last line of defense - though it may be perfectly correct - is a departure from the normal arrangement. Probably because in other situations, it's cheaper or more customary for humans to do an activity and then nonhuman backup is added, but planes are already expensive and complicated enough that turning them into flying robots is worth the cost.

Note also that there is no mention of a human supremacist argument in the way that someone might claim poetry or legal analysis must be human-driven. In this scenario, it seems as though the argument is not whether our planes should be flying robots, but whether the humans should be backup co-pilots or in-cockpit ground crew to the flying robots.

You are remarkably vague about these "activities that need to be backstopped by mechanical or computer safeguards", but I am guessing they are mostly activities where, if absolutely nothing happens, nobody gets hurt. Those are the sort of activities that lend themselves to automated "STOP EVERYTHING NOW!" safeguards.

Once an airplane is in the air - really, once it has accelerated to a sufficient velocity on the runway - a sequence of complex activities absolutely must occur in an unpredictable environment, on an unforgivable schedule, even if the hardware is broken. Or else everybody dies. Trained and experienced human beings are actually very good at this. Computers, are not. Computers do not exercise common sense or judgement, they do not think outside the box or improvise, and rather than being able to fix things that have gone wrong they turn out to be abysmally bad at even recognizing that things have gone wrong in the first place unless they go wrong in a simple and ordinary way.

As for "sleepy, lazy, intoxicated meatbags", speak for yourself, not the rest of the human race.

Which carmakers will take the Boeing route, and which the Airbus?

I'll guess Toyota will let the computer have final control, while Honda will let the human override. I hope the American carmakers stay out of the market for self-driving cars altogether; I don't trust the automated systems they already have.

As I think about it, the outcome with self-driving cars will likely be the result of the regulatory environment in each country. The U.S. will almost definitely require human override, but will everyone else follow? In the case where the highway system communicates with self-driving cars to improve traffic flow, will ultimate control rest with the highway control system? Which countries are most likely to have that sort of arrangement?

There are no commercial aircraft without pilots to perform such tasks as visual traffic avoidance. More importantly pilots have the legal (and often technical) ability to override the flight computers. Even drone aircraft generally have remote pilots.

In practice, I believe driverless car owners will retain the right to override their computers, and will certainly retain the ability in practice.

By doing all your due diligence when generating choices
for the aquaponics system, you may avoid making costly mistakes.
And you may know they come from the clean organic source.

Bacteria such as nitrobacter are designed for breaking down
the ammonia that's being continually produced by the fish.

Comments for this post are closed