Autopilot on Boeing and Airbus?

[anonymous-user]

djc206 said:

How much deviation is there at the moment? I know the PRNAV error is in theory fairly large but the reality is that aircraft fly very precise tracks. What sort of gains are you looking at and have you based your calculations on permitted deviation or radar derived/ aircraft data from operators?

Indeed.

Aircraft are already flying exactly along a track between points using multiple gps. I'm not sure what kind of optimisation there could be to improve this.

Hope it all works out anyway.

[rhinochopig]

Oli, are you familiar with IEC61508? Your team will probably need to be.

[Zad]

Conventional PID control systems are more than capable of keeping a vessel on whatever heading or course you want, and adapting to changing local conditions. Proving that such a system is stable under the full range of conditions is not simple, but it is only MSc level stuff.

e.g. http://uk.mathworks.com/help/robust/examples/tunin...

If you are looking at using measurement of atmospheric conditions in order to fly a predicted path, then you are up against a measurement problem as much as a maths prediction problem. Chaotic dynamic systems are a notoriously intractable problem, and CPU horsepower isn't the issue.

[anonymous-user]

I would also say that the biggest limitation to the dynamic path you can take is the Self Loading Freight in the back! They tend to get annoyed if you spill their drinks when chasing a thermal........ ;-)

[Olivero]

djc206 said:

How much deviation is there at the moment? I know the PRNAV error is in theory fairly large but the reality is that aircraft fly very precise tracks. What sort of gains are you looking at and have you based your calculations on permitted deviation or radar derived/ aircraft data from operators?

The deviation at the moment differs depending on aircraft type, weather and route. There is very little publicly available data to study for obvious reasons.
Gains would be from within the most narrow permitted deviation. The aim after all is not to change the overall flight route but rather optimize the path.

[Olivero]

rhinochopig said:

Oli, are you familiar with IEC61508? Your team will probably need to be.

Yes, very familiar...
Our system will not be mission critical. It would operate in addition to existing systems rather than any kind of replacement. A simple flip of a switch could turn it off without effecting autopilot.

[Olivero]

Zad said:

Chaotic dynamic systems are a notoriously intractable problem, and CPU horsepower isn't the issue.

This is our biggest strength. Already proven in a similar but different environment.

[djc206]

Olivero said:

The deviation at the moment differs depending on aircraft type, weather and route. There is very little publicly available data to study for obvious reasons.
Gains would be from within the most narrow permitted deviation. The aim after all is not to change the overall flight route but rather optimize the path.

Surely you need some data otherwise you've come up with a problem that as suggested by pilots above who reference the use of gps doesn't actually exist in the case of modern airliners? This would involve working with one of the major manufacturers or a major ANSP or ideally both?

[anonymous-user]

Olivero said:

rhinochopig said:

Oli, are you familiar with IEC61508? Your team will probably need to be.

Yes, very familiar...
Our system will not be mission critical. It would operate in addition to existing systems rather than any kind of replacement. A simple flip of a switch could turn it off without effecting autopilot.

The aircraft I fly has Rockwell Collins avionics, are you suggesting they would buy your system and integrate it with theirs or that each individual airline would buy it and retrofit it to the existing approved and certified avionics that come with the aircraft?

[AER]

I would reckon the biggest efficiency gains in air transport lie in managing traffic as it departs and converges on the airports. Modern airliners are at their most efficient in cruise, but waste most energy and time in the departure and arrival process. Your computing power could be better assigned to optimizing this problem which could include manipulating cruise speeds to program arrival to slot into a narrower arrival schedule window.

[Olivero]

AER said:

I would reckon the biggest efficiency gains in air transport lie in managing traffic as it departs and converges on the airports. Modern airliners are at their most efficient in cruise, but waste most energy and time in the departure and arrival process. Your computing power could be better assigned to optimizing this problem which could include manipulating cruise speeds to program arrival to slot into a narrower arrival schedule window.

This is part two of our 'solution'. We want to get the system up and running on a number of commercial aircraft first, then use individual data collected to construct testable hypotheses. It is a bit of a chicken and egg situation.

[Olivero]

el stovey said:

The aircraft I fly has Rockwell Collins avionics, are you suggesting they would buy your system and integrate it with theirs or that each individual airline would buy it and retrofit it to the existing approved and certified avionics that come with the aircraft?

Ideally this would be an add on to an existing system. We are not trying to 'reinvent the wheel' just improve it.
One limiting factor is the frequency of data collection in traditional systems.

[davepoth]

So super-local and directional weather forecasting along a flight track allowing for fuel burn optimisation that way?

I'm not sure whether the gains over the conventional method is going to be sufficient, especially since this is going to be a bit of a bandwidth hog.

Any thoughts about how much data will be going back and forth to the planes? That will be an ongoing cost, and would be quantifiable - you will be able to work out how much fuel you'll need to save to make it worthwhile quite easily, at least.

[AER]

This project looks like the sort of thing you could simulate in a network of X-plane (or similar) flight simulators. I'm sure the flight simmers have organized the modelling of their autopilots/FMS to be accurate enough for your purposes.

[nellyleelephant]

As someone who sort of works in the same business (military aircrew helmets, so aerospace) have you any idea how much trials will cost you?

I don't know the size of your business, but I would imagine that's going to be your biggest cost.

[Kempus]

Honestly not wanting to piss on your parade but the cost and testing involved to get the certification to make either a modification or install a new system on a commercial aircraft is going to be your biggest hurdle.

Instead look at your option 2. Airspace design is the biggest gripe for most airlines. Flying inefficient routes with turns instead of great circle tracks. We do collaborative decision making at some airports now where aircraft can only start at a certain time to prevent holding on the ground with engines running and making a hell of a lot of noise. Help solve these problems or reduce holding at major hubs in some way and I think you'd be on to a winner!

Edited by Kempus on Thursday 27th November 13:03

[trashbat]

Aside from the unlikely sounding nature of it all: have you considered what happens to your idea if for instance the FAA ever get around to implementing Free Flight, or Eurocontrol their version of the same?

[djc206]

AER said:

I would reckon the biggest efficiency gains in air transport lie in managing traffic as it departs and converges on the airports. Modern airliners are at their most efficient in cruise, but waste most energy and time in the departure and arrival process. Your computing power could be better assigned to optimizing this problem which could include manipulating cruise speeds to program arrival to slot into a narrower arrival schedule window.

This is already being worked on. At the moment aircraft are already being slowed down manually quite some distance out from heathrow in order to help reduce airborne holding. In future arrive not before times will be calculated by ground based systems using delay data from ANSP's.

[trashbat]

I worked on modelling some of that stuff, including free flight and conflict resolution, in 2004. It seems to have fallen off the radar (ho ho) somewhat since then. IIRC the FAA's twenty year plan from some time earlier said we'd be sharing airspace with UAVs, personal aircraft and spaceplanes by now.

Not for the first time do I get to ask: where the fk is my flying car?

[djc206]

Haha we've got enough in the skies without adding MLM's to our airways.

I think free flight within Europe is difficult because of the number of ANSP's with their own agendas. In Britain it's difficult because we built most of our airports within a 30 mile radius of london and we have a fairly substantial military requirement.

Conflict detection software is now very good, I don't think that's really the biggest obstacle.