Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-08 Thread Jim Wilson
David Megginson said:

 Ampere K. Hardraade wrote:
 
  One explanation for the violent maneuvers that I thought of is this: as the 
  distance between the plane and the waypoint decreases, the accuracy required 
  in the course calculations increases.  Since it takes time for the autopilot 
  to respond, and takes even more time for the plane itself to respond to the 
  commands of the autopilot, the plane will never align itself perfectly with 
  the waypoint.  Hence, the autopilot will keep trying to catch the waypoint 
  until the very last moment, thus causing the violent maneuvers.
 
 Yes, that happens with student pilots as well -- localizers, VORs, you name 
 it all become very sensitive as you get close.
 
  One solutions to the above problem is to pop the waypoint when the plane is 
  still some distance away, thereby preventing the autopilot from making all 
  those course adjustments.
 
 Or simply increase the allowable error at close distance, or use some other 
 kind of filtering function.
 

There is a look ahead filter in the current autopilot code.  It is being used
to anticipate ILS radial interception for the 747.  Maybe thrust control too.

Best,

Jim


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-05 Thread Durk Talsma
My observations are similar to this. However, I don't think that it's a 
problem with the accuracy of the calculations as such. The route manager code 
is relatively old and tries to set a straight course to it's next waypoint. 
It doesn't consider the wind blowing the aircraft off its lateral track. The 
autopilot tries to compensate by keeping the aircraft on a heading straight 
to the next waypoint. Now, as we get closer, this happens at an increasingly 
greater rate, until the rate of required course adjustment exceeds maximum 
turn-rate. I've seen a few occasions where the 747 couldn't reach the next 
waypoint because of this phenomenon and started flying around it in endless 
circles, until I manually popped it. 

Cheers,
Durk

On Saturday 05 June 2004 03:28, Ampere K. Hardraade wrote:
 The violent maneuvers I was describing occur when the plane is a few
 kilometers away from the waypoint.  Therefore, it should have little to do
 with the way that pid controller reacts to the jump in waypoints.

 One explanation for the violent maneuvers that I thought of is this: as the
 distance between the plane and the waypoint decreases, the accuracy
 required in the course calculations increases.  Since it takes time for the
 autopilot to respond, and takes even more time for the plane itself to
 respond to the commands of the autopilot, the plane will never align itself
 perfectly with the waypoint.  Hence, the autopilot will keep trying to
 catch the waypoint until the very last moment, thus causing the violent
 maneuvers.

 One solutions to the above problem is to pop the waypoint when the plane is
 still some distance away, thereby preventing the autopilot from making all
 those course adjustments.



___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-05 Thread Jorge Van Hemelryck
All these issues are a matter of adjusting the autopilot coefficients to
specific aircraft dynamic characteristics.

In order to avoid the spiral into waypoint problem, you should try and
implement something like this: compute a track to the waypoint from the
present position, memorize it, and subsequently correct the trajectory
so as to remain on that track (not heading). It's also better to correct
this by using a lateral error (distance to the memorized track, not only
angle), which makes these corrections independent from the distance to
the waypoint. The inputs are: aircraft track (not heading, or you have
to allow for a difference between held heading and desired track),
desired track, lateral error. The output is a rate of turn.

The pop waypoint condition is more efficient this way: aircraft beyond
a line orthogonal to the desired track to the waypoint, and located at a
distance 'd' from the waypoint (see diagram attached).

In real-life IFR, if you are flying a non-RNAV aircraft, you're actually
supposed to overfly the waypoints. I know, it sounded weird to me too.
Of course, no one will really scold you if you manage to nicely
anticipate the turn in order to find yourself just on track to the next
waypoint. RNAV aircraft are supposed to anticipate every turn so as not
to overshoot airways.

-- 
Jorge Van Hemelryck
attachment: nextwp.png___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread David Megginson
Ampere K. Hardraade wrote:
I think it will probably be a good idea to increase the distance that triggers 
the poping of waypoints.  Right now, the plane practically gets on top of the 
way point before switching to the next, which is causing some awfully violent 
maneuvers.
The best way to do this is to vary the distance based on an aircraft's 
airspeed, to give a rate-one turn or a roll of 20 degrees, whichever results 
in a shallower bank.

All the best,
David
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread Josh Babcock
Ampere K. Hardraade wrote:
I think it will probably be a good idea to increase the distance that triggers 
the poping of waypoints.  Right now, the plane practically gets on top of the 
way point before switching to the next, which is causing some awfully violent 
maneuvers.

Regards,
Ampere
On June 3, 2004 11:23 am, Durk Talsma wrote:
This is hardcoded in src/Autopilot/route_mgr.cxx, line 112:
   if ( wp_distance  200.0 ) {
   pop_waypoint();
distance is probably in meters, but I might be wrong here.
HTH,
Durk
On Wednesday 02 June 2004 21:57, Seamus Thomas Carroll wrote:
Hi,
Does anyone know where the variable is kept that holds the distance to
a waypoint that triggers moving the current waypoint to the next
waypoint?
Seamus

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
How do real life FMS's do it?  Does this generally vary?  If so maybe it would 
make sense to pick this number up from the autopilot XML with a default defined 
in preferences.xml.

Josh
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread David Megginson
Josh Babcock wrote:
How do real life FMS's do it?  Does this generally vary?  If so maybe it 
would make sense to pick this number up from the autopilot XML with a 
default defined in preferences.xml.
At any given speed, the same bank angle will give you the same turn rate 
(for a coordinated turn), no matter how big or small the plane is.  I don't 
know the proper equation, but the rule of thumb for a rate one turn (3 
degrees per second) is knots/10+7, so at 120 kt you will need (roughly) a 19 
degree bank for a rate-one turn; at 250 knots, you will need (roughly) a 32 
degree bank.

I don't think that fast passenger planes typically bank much more than 20 
degrees, though, so they will be doing a turn at considerably less than rate 
one.  Get the proper formulas, specify the maximum allowed bank angle, then 
crunch the numbers to see how far ahead the plane will have to start turning 
at its current airspeed to change course by a specified number of degrees.

All the best,
David
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread Lee Elliott
On Friday 04 June 2004 15:37, David Megginson wrote:
 Josh Babcock wrote:
  How do real life FMS's do it?  Does this generally vary?  If so maybe it
  would make sense to pick this number up from the autopilot XML with a
  default defined in preferences.xml.

 At any given speed, the same bank angle will give you the same turn rate
 (for a coordinated turn), no matter how big or small the plane is.  I don't
 know the proper equation, but the rule of thumb for a rate one turn (3
 degrees per second) is knots/10+7, so at 120 kt you will need (roughly) a
 19 degree bank for a rate-one turn; at 250 knots, you will need (roughly) a
 32 degree bank.

 I don't think that fast passenger planes typically bank much more than 20
 degrees, though, so they will be doing a turn at considerably less than
 rate one.  Get the proper formulas, specify the maximum allowed bank angle,
 then crunch the numbers to see how far ahead the plane will have to start
 turning at its current airspeed to change course by a specified number of
 degrees.


 All the best,


 David

If I understand the problem correctly, changing the distance at which the turn 
is initiated won't make any difference - it's the way that the pid controller 
reacts to the jump in input as the waypoint is popped that's causing the 
abrupt change.

For example, if there are two waypoints, directly north and south of each 
other, when one waypoint is popped the pid controller is presented with an 
input change of 180 deg, regardless of how far away from the waypoint it is.

LeeE

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread David Megginson
Lee Elliott wrote:
If I understand the problem correctly, changing the distance at which the turn 
is initiated won't make any difference - it's the way that the pid controller 
reacts to the jump in input as the waypoint is popped that's causing the 
abrupt change.

For example, if there are two waypoints, directly north and south of each 
other, when one waypoint is popped the pid controller is presented with an 
input change of 180 deg, regardless of how far away from the waypoint it is.
The 180-degree turn is a special problem no matter how you look at it -- the 
key is to make sure that the autopilot will not overbank the plane.

Otherwise, though, turn distance does matter -- if you switch to the next 
waypoint at the right moment, the plane will turn smoothly to its new track 
without overshooting the waypoint and possibly entering unprotected airspace.

All the best,
David
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread Lee Elliott
On Friday 04 June 2004 23:30, David Megginson wrote:
 Lee Elliott wrote:
  If I understand the problem correctly, changing the distance at which the
  turn is initiated won't make any difference - it's the way that the pid
  controller reacts to the jump in input as the waypoint is popped that's
  causing the abrupt change.
 
  For example, if there are two waypoints, directly north and south of each
  other, when one waypoint is popped the pid controller is presented with
  an input change of 180 deg, regardless of how far away from the waypoint
  it is.

 The 180-degree turn is a special problem no matter how you look at it --
 the key is to make sure that the autopilot will not overbank the plane.

 Otherwise, though, turn distance does matter -- if you switch to the next
 waypoint at the right moment, the plane will turn smoothly to its new track
 without overshooting the waypoint and possibly entering unprotected
 airspace.


 All the best,


 David

Ah - I don't think I understood the problem properly - I'll shut up.

LeeE

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-04 Thread Ampere K. Hardraade
The violent maneuvers I was describing occur when the plane is a few 
kilometers away from the waypoint.  Therefore, it should have little to do 
with the way that pid controller reacts to the jump in waypoints.

One explanation for the violent maneuvers that I thought of is this: as the 
distance between the plane and the waypoint decreases, the accuracy required 
in the course calculations increases.  Since it takes time for the autopilot 
to respond, and takes even more time for the plane itself to respond to the 
commands of the autopilot, the plane will never align itself perfectly with 
the waypoint.  Hence, the autopilot will keep trying to catch the waypoint 
until the very last moment, thus causing the violent maneuvers.

One solutions to the above problem is to pop the waypoint when the plane is 
still some distance away, thereby preventing the autopilot from making all 
those course adjustments.

I should have explained this more throughoutly.  Sorry for the inconvience.

Regards,
Ampere

On June 4, 2004 06:19 pm, Lee Elliott wrote:
 If I understand the problem correctly, changing the distance at which the
 turn is initiated won't make any difference - it's the way that the pid
 controller reacts to the jump in input as the waypoint is popped that's
 causing the abrupt change.

 For example, if there are two waypoints, directly north and south of each
 other, when one waypoint is popped the pid controller is presented with an
 input change of 180 deg, regardless of how far away from the waypoint it
 is.

 LeeE

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-03 Thread Durk Talsma
This is hardcoded in src/Autopilot/route_mgr.cxx, line 112:

if ( wp_distance  200.0 ) {
pop_waypoint();

distance is probably in meters, but I might be wrong here.

HTH,
Durk

On Wednesday 02 June 2004 21:57, Seamus Thomas Carroll wrote:
 Hi,

 Does anyone know where the variable is kept that holds the distance to
 a waypoint that triggers moving the current waypoint to the next waypoint?

 Seamus



 ___
 Flightgear-devel mailing list
 [EMAIL PROTECTED]
 http://mail.flightgear.org/mailman/listinfo/flightgear-devel


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel


Re: [Flightgear-devel] How to change minimum distance to activate next waypoint?

2004-06-03 Thread Ampere K. Hardraade
I think it will probably be a good idea to increase the distance that triggers 
the poping of waypoints.  Right now, the plane practically gets on top of the 
way point before switching to the next, which is causing some awfully violent 
maneuvers.

Regards,
Ampere

On June 3, 2004 11:23 am, Durk Talsma wrote:
 This is hardcoded in src/Autopilot/route_mgr.cxx, line 112:

 if ( wp_distance  200.0 ) {
 pop_waypoint();

 distance is probably in meters, but I might be wrong here.

 HTH,
 Durk

 On Wednesday 02 June 2004 21:57, Seamus Thomas Carroll wrote:
  Hi,
 
  Does anyone know where the variable is kept that holds the distance to
  a waypoint that triggers moving the current waypoint to the next
  waypoint?
 
  Seamus
 
 
 
  ___
  Flightgear-devel mailing list
  [EMAIL PROTECTED]
  http://mail.flightgear.org/mailman/listinfo/flightgear-devel

 ___
 Flightgear-devel mailing list
 [EMAIL PROTECTED]
 http://mail.flightgear.org/mailman/listinfo/flightgear-devel

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel