Virtual Pilot Lands Qantas Jet

An anonymous reader writes "Australian airline Qantas has successfully tested an automated landing where both the pilot and the control tower didn't talk to each other. The plane was being piloted by a "Virtual Pilot" located in the control tower."

Comments

In Related News...

[fuzzybunny]

...a Quantas flight carrying 357 passengers and crew plummeted to its destruction for unknown reasons.

Sydney air traffic control reports picking up garbled radio traffic fragments, but is still trying to decode the meaning of "D00D! U G0T PWNT!!!"

What about the pilot jobs?

[Thinkit4]

We really have to think of alternate economic systems the more stories like this come up. An automated car doesn't really displace jobs--but this really could.

Re:What about the pilot jobs?

[pwagland]

We really have to think of alternate economic systems the more stories like this come up. An automated car doesn't really displace jobs--but this really could.

I think you say that without realising how many taxi drivers there are in the world. I would be prepared to bet that the number of taxi drivers compared to the number of qualified superliner pilots would be quite large...

Food for thought.

Re:What about the pilot jobs?

[Loadmaster]

One thing that will keep these from replacing a human is emergency procedures. Even with redundant systems something could go wrong that this computer can't fix.

Also, at least for the time being, most people wouldn't trust a computer to fly them anywhere. It's true that autopilot can do a lot, but most people don't think of it because they know there's a human being at the controls.

Remote control of air planes

[bomblaster]

Airplanes have been able to land on auto pilot for years using the Instrument Landing System (ILS)!!

This is more about remote control of an air plane than automated landing. According to the article, digital commands were uploaded to the 747.

With all this technology already in place, it is certainly possible to develop systems to enable commercial air planes take off on auto-pilot too. But that will require huge costs in new infrastructure to be installed at airports similar to the ILS for landing. Real-time software testing costs will also be enormous. Maybe FedEx mighe be interested in funding this :-)

Re:Remote control of air planes

[Muad'Dave]

I heard that autopilot-controlled ILS landings were routine for cargo carriers, but the FAA doesn't allow it when passengers are aboard. UPS and FedEx can do zero by zero landings (no visibility at all) if the controllers allow it.

Re:Remote control of air planes

[grozzie2]

I heard that autopilot-controlled ILS landings were routine for cargo carriers, but the FAA doesn't allow it when passengers are aboard.

you heard wrong. The rules are very specific in this area. A 'normal' instrument landing into weather of cielings greater than 200 feet with visibility greater than 1/2 mile may be flown by a pilot, or by the autoland system, with no preference. A Category II ILS (100 feet, 1/4 mile) _should_ be flown by the autoland, but pilots may elect to hand fly if equipment problems dictate so. A Category III ils (cieling 0, visibility 0) may NOT be hand flown and MUST be flown by autoland, no exceptions. A non functional autoland is cause to divert to an alternate airport with weather conditions suitable for a hand flown landing.

Fedex and UPS do a lot more zero-zero autoland operations because they have invested in the equipment to allow them to do so on most of thier fleets. Very few airlines actually invest in the equipment and training to do CAT III approaches, and it's surprising how many dont even invest in the requirements for CAT II.

From a safety perspective, this is an absolute no brainer. Just look at the statistics, and look for the number of landing accidents attributed to 'pilot error', and then look for the number attributed to 'autoland malfunction'. Ratio those numbers against the number of hand flown landings, vs the number of autolands performed over the sample period. You'll find the difference between the ratios to be larger than 3 orders of magnitude, with the autoland a clear winner.

When it gets right down to it, in normal operations, on autoland capable equipment, there is really no reason to hand fly airplanes today. The autoland will do a safer job, and pilots get plenty of practise in the 'bad situations' in the simulator. The flight management system will fly a more precise profile, using less fuel, with a higher safety factor. This translates into safer operations and lower costs.

I've been flying for 28 years, made my living flying airplanes for 18 of those 28. Pilots of large aircraft today are nothing more than systems managers. They assemble and analyze data in real time, and act as keypunch operators for the flight management system. They provide an audio interface between the air traffic control system and the flight managment system keyboard. The article was completely out of context of reality, in reality the system being tested just eliminates the audio step of transferring air traffic control clearances into the on board flight management system, and instead transfers them via a data link, thereby removing the potential for keypunch errors.

The reporter writing the article saw something they didn't understand, and tried to dramatize it to a point it's totally removed from reality. There was no remote control, just automated data transfers of data that would normally be spoken then keypunched. I guess the reporter must be a /. regular....

Is this really big news?

[kabocox]

I know the mil. has had auto landing and take off tech for years according to popular science. I know com. airlines have autopilot on most big planes. I just figured that it there was more of a political reason than a tech. reason why pilots haven't been entirely automated. I'd rather have a trained human "flying" my aircraft, but it may be faster/safer/cheaper to have a computer do it. The big reason that I've heard that we will always need pilots is if anything happened midair the pilot could either fix it fly around it recongizing that the incoming data from his instruments couldn't possibly be true.

Actually, I've always wanted an autopilot for my car. I'd feel alot safer if there was a dependable/safe/cheap autopilot for cars. Most car accidents are caused by human error. I'd love to prevent human error from my car.

Re:Is this really big news?

[parksie]

Commercially, the autopilot is the preferred method because it can make the tiny imperceptible changes for maximum fuel efficiency, that a normal pilot wouldn't (they're unnecessary from a flight point of view).

When the autopilot fails, the normal pilot takes over.

Re:Is this really big news?

[Anonymous Coward]

Once I had a lay over, so I spent the night at one of those off-airport hotels where aircrews rest overnight.

I went down to the bar expecting the SAS bikini team to be there, only to find a lone senior pilot. Anyhow, there was nothing else to do around there, so we chatted over a beer while watching the game. Among the things he said were: "9 out of 10 landings are by autopilot. The tenth one has to be manual according to the book so we don't forget how to land manually.... You know how sometimes the landing is quite bumpy? that's the manual one."

Re:Is this really big news?

[SuperBanana]

I just figured that it there was more of a political reason than a tech. reason why pilots haven't been entirely automated.

In the case of landings, it's called "weather", or more specifically, "wind". My father flew single-engine planes and you had to have lightning fast reflexes on windy days. It's better in a jumbo because of their mass, but they get tossed around a decent amount too.

Pilots are in the cockpit because planes are complicated. Planes can, and often do, suffer from mechanical breakdown, requiring improvisation or "best effort" solutions. Instruments do regularly fail requiring again, educated guessing or getting the information in other ways; GPS fails? Pilot whips out the charts and uses the ADF to navigate by VHF beacons. Avionics fail? Well, landmark time. The pilot can handle it- the computer goes "beep" and prints out "service code #432565, call Boeing". You rave about autopilots, but we've had a string of problems with both autopilots and some control systems- most infamously the Airbus disaster where the plane gave the pilot the wrong altitude AND refused to let him apply full power to save the plane from crashing.

We have this redundant setup whereby the actions of the pilot in command are confirmed by the copilot. And if one kicks the bucket entirely, the other one is fully capable of taking over. They've even often got a guy just to handle watching over the plane(engineer) in many cases.

Most major incidences in the last 10 years of US commercial airflight were due to mechanical failure. Not pilot error. The very last fucking thing we need to do is put more machines into the equation when we've proven we can't handle building them correctly in the first place.

Landing was tough

[4of12]

Airplanes have been able to land on auto pilot for years using the Instrument Landing System (ILS)!!

Interesting. I don't keep up with this technology, but years ago the landing was the diciest part of flying a plane which couldn't be automated, where cruising and take-offs could be automated.

Even now, when I'm on a commercial flight, I always implicitly rate my pilot based on the landing, how much of a bump, whether we hop, etc. Just like I rate Chinese restaurants based on their Hot & Sour Soup.

Re:Virtual Terrorist?

[WayneConrad]

How long until a virtual terrorist hijacks the uplink and "lands" this automated plane in a building?

I'll bet that nobody thought of that. We know that aviation people just don't give a darn for safety. And people just don't worry about protecting themselves from terrible things that have already happened.

There's probably not a pilot on the plane who can take over any time he wants.

The whole system is probably connected to the internet, too.

Heck, the uplink is probably unencrypted tones on a CB channel that any kid could generate by whistling into his walkie talkie.

Yes, we should be scared.

Re:I think this is probably driven by...

[rkhalloran]

Thinking of ST II: Wrath of Khan, where Kirk uses the 'prefix code' of the ship Khan's hijacked to drop its shields just before blasting it... Then you have the problem of keeping the in-cabin override switch unknown to the general public so the hijackers can't just lock out the tower as soon as they take the cockpit.

Coming Soon

[mrlsd]

Coming soon: virtual passengers to save the airline industry.