paparazzi-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Paparazzi-devel] barometer or not?


From: Vassilis V.
Subject: Re: [Paparazzi-devel] barometer or not?
Date: Mon, 2 Nov 2009 21:49:39 -0500

Joe,


updates ( ie, the pitch from altitude loop). We have modified the Pap. Way
of doing things considerably and do not control altitude via airspeed, but
have adopted more conventional control, with elevator from pitch command,
aileron from roll command,  throttle from airspeed command as our inner
loops.

The same approach is used in the latest Paparazzi code when using the airspeed sensor (i.e. USE_AIRSPEED defined).


I am not sure what you mean by

"to get the full benefits of  
the barometric sensor we should update at 10Hz or more"

Do you mean to update the inner loops at 10 Hz or just sample the sensor and
compute the result (Alt) at 10Hz, ensuring that when the inner loop command
update occurs, that it is using the very latest sensor status (a latency of
100ms max)? Updating the inner loops at 10Hz would seem excessive, as the
elevator, etc, could hardly be expected to respond much in 100ms, in fact,
most cheap analog servos are as slow as this anyway.

I am referring to the frequency of the vertical control loops (v_ctl_altitude_loop and v_ctl_climb_auto_throttle_loop), which is currently 4Hz. The roll and pitch inner loops (h_ctl_attitude_loop) run at 60Hz as far as I can tell so the servos should be updated approximately as fast. What I believe that might improve performance is to run the barometer and vertical control loops faster. Of course at this point is only a theory since I have not yet tested it. I am also getting very good performance even at 4Hz vertical control loop update (and 20Hz barometer update), my Easystar stays about 2-3 meters from the desired altitude.



We have achieved very good flight results, with altitude control within 3 to
4 meters of setpoint, flying at 22m/s, with wind speed of 8 to 11m/s.
Horizontal control is equally good, with circles and figure eight patterns
yielding track errors of less than 3meters in similar winds. We do auto
take-of from standstill and it works perfectly. We are now working on the
autoland, using ultrasound sensors for the AGL sensing, and the autoland
sequence (spiral till 50m AGL, speed = 20m/s, the exit on the spiral tangent
to the specified touchdown point, as airspeed of 18m/s, then kill throttle
at 4m AGL and proceed to flare point at touchdown) works very well - have
done many 'virtual landings' at 20meters up, just need good AGL sensing! I
have, at the virtual touchdown point, switched to manual (throttle closed)
and just held the A/C till the flare point and touched down with ease, so
soon-soon!

Anyway, Baro-Alt is the way to go; 200ms for alt update is not great, and
who know what data lag is in the GPS internal computation of the Alt
anyway..

Have fun

Joe

-----Original Message-----
From: address@hidden
[mailto:address@hidden
] On Behalf Of Vassilis V.
Sent: Sunday, November 01, 2009 9:49 PM
To: address@hidden
Subject: Re: [Paparazzi-devel] barometer or not?

My 2c for the barometer, I used the ETS sensor and not the SCP1000 but  
they should behave similarly.
- You will have to update the Kalman parameters very likely, the ones  
used for the SCP1000 did not work for me (too much filtering and lag).  
I updated SVN with what currently works fairly well for my setup (see  
R and SIGMA2 constants in baro_ets.c).
- The altitude loops are updated at 4Hz, to get the full benefits of  
the barometric sensor we should update at 10Hz or more. I have not  
made that change yet in the code, I am still running at 4Hz.

Vassilis


On Nov 1, 2009, at 12:19 PM, antoine drouin wrote:

On Sun, Nov 1, 2009 at 6:04 PM, Chris <address@hidden> wrote:
I just finished rewriting all the SCP1000 files including the use  
of the
barometer as an alternative altimeter for the autopilot.
The files affected are the baro_scp.x ,estimator.x and main_ap.x
In simulation it works fine but i was wondering if the reading of  
this
barometer are more accurate than the gps altitude especially know  
with EGNOS
support.
Any opinions?

It's not so much about accuracy than latency. GPS is filtered, so
lagging badly and also low bandwidth.
You have some GPS/baro data recorded from a couple of years ago in the
svn in the sw/in_progress/barometer directory if you wanna plot or
play with filters

I am also writing some new files for the  SRF08  for use in auto  
landings
now that after my heart problems i got me a nice two months leave.

I never got anything interesting from a ultrasonic sensor on fast
moving vehicles  (read fixed wing)

Regards

Poine


_______________________________________________
Paparazzi-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/paparazzi-devel



_______________________________________________
Paparazzi-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/paparazzi-devel



_______________________________________________
Paparazzi-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/paparazzi-devel


reply via email to

[Prev in Thread] Current Thread [Next in Thread]