scrubbedexpat099

The ultimate test of AI would be their own discussion boards and Off Topic section.

Michael

That is one of the major problems with AI. One of the conclusions from the investigation was that since many new commercial airplanes can fly itself from take off to landing without pilot intervention, pilots are not trained when abnormalities occurred. Some assume that if the airplane was 25 years old, it is likely that the crash wouldn't have occurred since the pilots would have been trained to handle such an occurrence.

steveq

I don't think it helped that the senior pilot was asleep on a break either.

Michael

That is another problem with AI. With the plane having the ability to fly itself and collusion avoidance enabled, the pilot probably assumed that the airplane could fly itself so he probably wasn't very alert for any possible problems.

steveq

An interesting point. I wonder IF the plane could have corrected the problem, if the pilots hadn't intervened. I find it hard to believe the SOLE air speed indication system was the pitot tubes.

Michael

The auto pilot disengaged itself so the pilots had to fly the plane. Without correct air speed indicators and no horizon since it was night time, it is hard for a pilot to figure out whether the plane is climbing, falling, or flying straight. Apparently the pilot thought the plane was falling because of the low air speed indications, was completely disoriented, and did exactly the opposite of what he should have done. I suspect he believed that there wasn't enough thrust from the engines and assumed that was causing the low air speed and the plane to fall.

steveq

They'd've had artificial horizons though. Did they assume they weren't working either ?

Michael

I don't think the artificial horizons will tell the pilot whether the plane had previously climbed or fallen but just that the plane is flying level with the horizon.

There are a lot of unanswered questions concerning the pilots state of mind. Apparently for some reason (from the voice recordings), it appears that he believed he was flying very low just above the ocean but he must have ignored or didn't believe the altitude indicator.

Like many airplane accidents, when one indicator is inaccurate, pilots aren't sure whether other indicators are working properly or not. That is the same reason that the auto pilot disengaged since the auto pilot is not sure how to properly fly the plane with an inaccurate air speed indicator and assumes a human will analyze the problem correctly allowing the plane to fly to it's destination.

Zen10

I haven't followed this case at all - is there a reason why ATC couldn't give him his altitude from radar?

Michael

They never called the tower to notify them that they had a problem and they were not in range of any ATC radar to see them with reflective information.

Even for the tower to get transponder information (the airplane's controls responding with the information), the plane has to be pretty close to the ATC tower.

Although Flight 447 had an Addressing and Reporting System (ACARS) but that is automatic and the plane kept flying after the last transmission so it was difficult to find the wreckage.

http://www.cnn.com/2009/TECH/06/03/d....nav.tracking/

steveq

Would GPS have given them enough information, I know its only "good" to 10 metres or so....but in the many thousands they were above the atlantic when this all kicked off....

Maybe they distrusted ALL their sensors and relied on their senses, which were wrong.

civilservant

Woud rather they were at the controls dog tired because its a long flight?

The problem with automation is that many pilots are using it from the second they leave the runway to the second they touch down. They never hand fly the aircraft. The A330 is designed to not allow a pilot to stal the aircraft, it has limitations on Angle of Attack etc to prevent it.

The automation failed because the pilot static system (a small tube outside the aircraft) iced up. So these millions of pounds of computer failed because of some ice, because they were received erratic information and the autopilot disconnected.

I recognise that in this situation the pilots failed to respond appropriately, but what if this had been an unmanned aircraft? There would have been no chance at all.

Over the ocean, at night, there are no visual references. The brain cannot handle the lack of input an you become disorientated. There are cases of pilots entering clouds and exiting them upside down because they became completely disorientated.

There was a recent case of a Spanish airline that was nearly a fatal accident. There was a last minute change of runway, and rather than fly the plane onto the damn ground the captain was caught dialling the new information into the Flight Management Computer. The airlines themselves have caused this, because of their instructions to pilots to use the FMS as much as possible. It saves fuel, the problem is it dosent allow pilots to keep their hand flying skills current.

I am looking at doing my commercial pilots license, and this is one thing that genuinely concerns me. In a little Cessna 152 its not a problem because there isn't any.

When an autopilot detects conflicting information from the P/S system it is designed to disconnect to allow the pilots to address the situation.

There's an old saying - Aviate, Navigate, Communicate. Fly the plane, figure out where you are, tell someone the problem.

They failed the first one. Its not the first time a pilot has stalled a perfectly good aircraft. Colgan Air 3407. Birgenair 301. Aeroperu 603 was very similar to AF447.

steveq

I thought they were only 2-3 hours out ?

If it were an unmanned plane, it would have had a better armoury of sensors.
Surely the thing ALSO still has an inertial system, as well as GPS ?

civilservant

At the time yes, however it was to be a long flight, and the Pilot in Command of long distance flights is usually in the left seat during the critical phases of flight.

i.e Take-Off, Climb out and onto route, then decent and landing.

It should be noted that the flight time of 13 hours exceeds the maximum duty day allowed by Air France of 10 hours.

Michael

GPS can't indicate altitude but only coordinates. Even planes that are equipped with ACARS using a SATCON datalink such as flight 447 normally is not monitored in real time. If it was monitored in real time, Air France would have known that the flight 447 was having problems and may have tried to contact the pilot using HF (not that reliable). Instead Air France didn't even suspect problems until after the pilot didn't make an expected voice communications.

There is a proposal for a new ATC system where there will be towers throughout the continents that will constantly receive all the airplane's indicators (GPS, altitude, air speed, etc.) which can be fed to ATC centers instead of passing planes off between ATC centers which currently uses using radar and transponders for tracking. ATC centers still will need radar since many light planes probably won't have the equipment. However that concept won't work over oceans unless they can acquire enough bandwidth and satellite capability to handle constant transmissions from all the planes over the oceans at the same time.

I suspect the proposal to use towers instead of satellites is because of the lack of bandwidth available. By using towers the amount of bandwidth needed will be greatly reduced since transmissions will be line of sight (maybe 100 miles) and all planes can probably share the same frequency band since only a limited number of planes will be within the range of any receiving tower.