Board index FlightGear Development Aircraft Autopilot and route manager

IT-Autoflight or generic FG AP problem?

Designing a stable autopilot is one of the hardest things. Need help?

IT-Autoflight or generic FG AP problem?

Postby Darkriser » Fri Jan 19, 2018 3:39 pm

Recently, I managed to obtain Level-D format of AIRAC data from Navigraph and I tell you,
flying real SID/STAR/Approach procedures takes the experience to a totally different level!

However, I noticed few problems when flying AP, but I don't know whether they are IT-Autoflight-related
or it's a generic FG Autopilot problem (I'm rarely flying other than IDG airplanes).

Let's have an instrument approach LZTT ILS RWY27 (page 7).

  1. Can't remember the A32X, but I'm sure that PA28-Warrior did NOT fly over TP after a missed approach,
    turned left prematurely and continued to PPD. I've actually verified that TP is marked as "fly-over"
    in the procedures xml file. Why the short-cut then?
  2. The missed approach procedure ends in a hold over ABRAG. Neither A32X nor PA28 were able to enter
    the hold (although it was correctly drawn on the Equipment->Map and A32X's ND. After reaching ABRAG
    there was no active waypoint in Route Manager and in both cases the planes were heading SW.
Are my expectations overestimated, is there a generic FG AP problem or is this an IT-A related issue?
Many thanks for your opinions/responses...
Darkriser
 
Posts: 64
Joined: Wed Dec 28, 2016 12:27 pm
Location: LZIB
Version: next
OS: Arch Linux

Re: IT-Autoflight or generic FG AP problem?

Postby V12 » Fri Jan 19, 2018 4:02 pm

Darkriser wrote in Fri Jan 19, 2018 3:39 pm:...flying real SID/STAR/Approach procedures takes the experience to a totally different level!...

Yes, this is very very true ! New dimension of the flying with FG.
Fly high, fly fast - fly Concorde !
User avatar
V12
 
Posts: 1570
Joined: Thu Jan 12, 2017 4:27 pm
Location: LZIB
Callsign: BAWV12

Re: IT-Autoflight or generic FG AP problem?

Postby Octal450 » Fri Jan 19, 2018 4:33 pm

IT-AUTOFLIGHT does not support holding patterns as of now. Support it planned to be added only to:
* IDG-A32X
* IDG-A33X
* IDG-MD-11X

I have no plans to bring that to the core IT-AUTOFLIGHT at this time.

As for actually flying the SID/STAR, Wecsje does it all the time on VATSIM, as IT-AUTOFLIGHT now has predictive turning, and takes crosswinds into account. So, they fly relatively well.

Keep in mind that while IT-AUTOFLIGHT is 100% custom, it still use the Route Manager for route management. The course calculation is custom. The planes mentioned above will have a custom route management, so that problem will be solved, again, I don't plan to backport that to the core as of now.

Kind Regards,
Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: IT-Autoflight or generic FG AP problem?

Postby Darkriser » Fri Jan 19, 2018 9:23 pm

it0uchpods wrote in Fri Jan 19, 2018 4:33 pm:...planned to be added only to:...

This will be absolutely sufficient for my needs :-)

But what about the 'fly-over' waypoint?
Is it possible that your course calculations may ignore the 'fly-over' property?
I can confirm this behavior for PA28 (I guess it also uses IT-A), but will verify for the bigger beauties (A's and MD's).
Darkriser
 
Posts: 64
Joined: Wed Dec 28, 2016 12:27 pm
Location: LZIB
Version: next
OS: Arch Linux

Re: IT-Autoflight or generic FG AP problem?

Postby Octal450 » Fri Jan 19, 2018 11:52 pm

I believe it does, I can check that out.

Kind Regards,
Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: IT-Autoflight or generic FG AP problem?

Postby Darkriser » Sat Jan 20, 2018 2:05 pm

it0uchpods wrote in Fri Jan 19, 2018 11:52 pm:I believe it does, I can check that out.

Please do, I verified and confirmed that behaviour also with your A320.
This is my missed approach procedure I performed today (same chart as in my orig post at the top):
After a MA the procedure says to climb and maintain RWY heading until D271J and then turn left back to PPD.
However, IT-A realizes that it's easier to turn back immediately even though D271J is designed as 'fly-over' waypoint.

Another problem:
Exactly the same (premature turns) happens for 'fly-by' waypoints. Is there (IRL) any requirement for a max distance from a fly-by
waypoint where an airplane must get in order to mark it as passed?
I can provide a screenshot, but look at the approach plate from my orig post above (STAR on pg. 5 and approach on pg. 7).

Navigraph's data almost perfectly copy this approach:
after you pass EPEDA, continue to PPD, ABRAG, then there's a fly-by waypoint at 9.9 DME PPD, turn back to ABRAG and track ILS.

However, after passing PPD and ABRAG, IT-A realizes that 9.9 DME PPD wpy is useles, because the next wpy is ABRAG again.
So it skips that wpy and turns ABRAG as the next wpy.
However, we're still very close to ABRAG so IT-A also marks it as passed and turns the next wpy LZTT RWY 27 ILS as active.
The result (very unrealistic) is that the airplane turns around immediately after ABRAG and takes course directly to LZTT.
I know that 9.9 DME PPD waypoint is marked as fly-by, but IMHO the plane should fly reasonably close before turning to next wpt,
shouldn't it??

Josh (and other guys at IDG) , don't get me wrong. I know these are very marginal situations. But I consider your planes for
almost perfect (at least for my needs) and would like to make them completely perfect (if possible, of course) by reporting such
marginal issues...
Darkriser
 
Posts: 64
Joined: Wed Dec 28, 2016 12:27 pm
Location: LZIB
Version: next
OS: Arch Linux

Re: IT-Autoflight or generic FG AP problem?

Postby Octal450 » Sat Jan 20, 2018 2:40 pm

Hi,
Don't worry. It's on the list of things to do, but we are waiting for the FMGS/MCDU work to further, as there is a FLYOVER key on the MCDU, where you can tell it to fly-over.

In the mean time if I get spare time, I will see if I can get a quick and dirty fix to check this, providing that the Route Manager knows fly-overs. If it does, I can add that easily to IT-A and thats all thats needed.

It will be added, but not quite yet, as there are more important things at the moment to do.

Thanks and Kind Regards,
Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: IT-Autoflight or generic FG AP problem?

Postby Wecsje » Sun Jan 21, 2018 11:02 pm

The routemanager knows nothing... Yes, it can read simple sid's and star's from level d data. But it doesn't recognize transitions, and on an airport like EBBR it decides to just not work completely
Twitch Streams: https://www.twitch.tv/wecsjelive
Contact methods: Discord (Wecsje#6351), FlightSims United discord (https://discord.me/flightsimsunited), Steam (Wecsje)
Track me on VATSIM: https://vatstats.net/pilots/1397313
User avatar
Wecsje
 
Posts: 204
Joined: Sun Feb 09, 2014 10:13 am
Location: The Closet, Under the Stairs, the Netherlands
Callsign: WS208J
Version: Newest
OS: Windows 10 Pro

Re: IT-Autoflight or generic FG AP problem?

Postby Octal450 » Sun Jan 21, 2018 11:22 pm

We will see if it does, it might just not have the GUI to support using it.

I'll see what I can do about it.

Kind Regards,
Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: IT-Autoflight or generic FG AP problem?

Postby Isaak » Tue Jan 23, 2018 12:50 am

Wecsje wrote in Sun Jan 21, 2018 11:02 pm:The routemanager knows nothing... Yes, it can read simple sid's and star's from level d data. But it doesn't recognize transitions, and on an airport like EBBR it decides to just not work completely


At EBBR it doesn't work because runway 02/20 is now 01/19 in reality and thus in the navigraph file. You have to either modify your apt.dat (replacing EBBR lines with current rwy layout) or modify your EBBR.procedures.xml so that every instance of rwy 01 or rwy 19 is replaced with 02 or 20.

This is the same for any airport which has changed/added/removed runways or rwy numbers.
Isaak
 
Posts: 467
Joined: Sat Jun 04, 2011 2:52 pm
Location: Leuven, Belgium
Callsign: OO-ISA
Version: 2019.1.1
OS: Windows 10

Re: IT-Autoflight or generic FG AP problem?

Postby zakalawe » Sun Jan 28, 2018 2:36 pm

Hi, Stuart, kindly pointed me at this thread, just a few points:

- the flight-plan system (underneath the route manager) knows fly-by vs fly-over, read from the Level-D data

- it also knows about transitions, if you enter an enroute point in the route manager and then select a SID or STAR, it will pick the matching transition automatically if one exists. if the first/last enroute point is not a transition point, we use the closest transition point I think. (Will need to check - there's probably bugs to be found in this logic as well)

- there's unfortunately nothing we can do if the data for procedures doesn't match our runway definitions - in that case (such as EBBR) you have to fix the procedures XML data until we update the main nav DB I'm afraid :(

For the default 'GPS' which is what most people mean by the autopilot, which tries to fly the route, turn anticipation should arm within 2nm of the waypoint for fly-by waypoints, and the WP should sequence when passing a line where the bearing the to the WP is perpendicular to the direction of flight. I.e exactly half-way through the turn when turn-anticipation is active.

(The 2nm distance can be configured by aircraft in the GPS /config/ property section)

Of course if the aircraft is suing a custom GPS/FMS implementation the turn-arm, waypoint sequencing and turn-anticipation logic will be different.

BTW, it's always easy to add additional FlightPlan APis in Nasal if requested, eg to manually select the transition or to toggle the WP fly-by / fly-over type (actually this one already exists), just send a request to the development list and ideally some test data / test case using the UFO or default AP.

Cheers :)
zakalawe
 
Posts: 1152
Joined: Sat Jul 19, 2008 4:48 pm
Location: Edinburgh, Scotland
Callsign: G-ZKLW
Version: next
OS: Mac

Re: IT-Autoflight or generic FG AP problem?

Postby Octal450 » Sun Jan 28, 2018 3:42 pm

Hi,
We use non default AP,
The course has corrections for drift and such, and it calculates turn anticipation so it gets in the line for the next waypoint. Its working very well. Good to know the underlying system is supporting it. That means when the custom nasal flight plan is released, and the route manager is removed, it will support it.

Kind Regards,
Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: IT-Autoflight or generic FG AP problem?

Postby V12 » Mon Jan 29, 2018 9:14 am

I made 2 test flights with A320 :
- from LOWI 26 SID RTT2X to LIPB 01 STAR FORE1B, flight log at https://fgtracker.ml/modules/fgtracker/ ... ID=6984439
- from LIPB 19 SID ALESE5A to LOWI 26 approach LOC RWY26, flight log at https://fgtracker.ml/modules/fgtracker/ ... ID=6984464

The plane follow route as precise, as speed allowed it. Departure arc over RTT is very wide, because 300 kts TAS. All other turns was very precise. In that cases route follow worked perfectly.

For better adventure I tested LOWI 26 SID RTT2X with Concorde. The plane was able follow the route at 230 kts TAS (Airbus only 190 kts TAS), but with average G-load 1.3g (flight log here - https://fgtracker.ml/modules/fgtracker/ ... ID=6984479)
Fly high, fly fast - fly Concorde !
User avatar
V12
 
Posts: 1570
Joined: Thu Jan 12, 2017 4:27 pm
Location: LZIB
Callsign: BAWV12


Return to Autopilot and route manager

Who is online

Users browsing this forum: Applebot [Bot] and 1 guest