RE: [Autopilot] The simulator versus real life

[Kahn, A. -A. <Aar...@it...>]

Hello,
  What you have found is what I would expect to see.  You see, the simulator
is not modelling all of the linkages and levers that are present in the
helicopter.  The simulator is assuming all of this has been accounted for
already, so the commands from the controller are direct angles on the
swashplate and tail rotor.
  In reality, though, we have not recreated this quite perfectly.  The way
that the controller we flew before is setup with the following block flow
chart...

command-->0-K-> angle2servos -> linkages2angles ---> swashplate ---> heli
        - |                                                            |
          +------------------------------------------------------------+

As you can see, there are a few things in the loop after the gains and
before the swashplate.  All of theses blocks act as gains.  The
linkages2angles can be represensted by Kl and linkages2angles can be
represented by Ka.  The resulting gain in the control loop then is K*Kl*Ka.
Now if there is an error in Kl modelling, this can be made up in the K
value.
   Also, there are inherent modelling errors in the simulation as compared
to real life.  To reduce these errors, one needs to do a full system
identification on the aircraft.  This would be used to more accurately
determine the parameters of the math model.
   A first step on this road of parameter identification would be to very
accurately measure all of the linkage gains, mass properties, and
dimentional properties of the helicopter.

I hope that this might shed some light on the subject.

Take Care,
Aaron Kahn

-----Original Message-----
From: Tramm Hudson
To: aut...@li...
Sent: 9/29/2002 10:43 AM
Subject: [Autopilot] The simulator versus real life

I've hooked up the flight controller that we've used for many
successfull
real flights to the simulator and found that it doesn't work as well
as before...

The gains that we had previously were rock solid for the roll axis
and pretty good for the pitch.  The roll axis is still decent, but the
pitch axis has quite a bit of oscillation.

We have increased the proportional term of the roll controller slightly,
decreased the proportional pitch quite a bit and zeroed the derivative
for pitch.  If I add the derivative term back in, the pitch axis
smooths out a bit but still has small oscillations.

Currently all of these values are hard coded into the source code.
We really should have a configuration file that specifies the gains,
the radio channels and the IMU arrangement.  It would have to be
both a compile time and a runtime thing -- the IMU board would need
to have these compiled in for safety, while the flight controller
gains would be runtime and even changable in flight.

Does anyone want to tackle this job?

Trammell
-- 
  -----|----- hu...@sw...                     H 240-476-1373
*>=====[]L\   hu...@ro...               W 240-453-3317
'     -'-`-   http://www.swcp.com/~hudson/                KC5RNF



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
http://autopilot.sourceforge.net/     Development mailing list
https://lists.sourceforge.net/lists/listinfo/autopilot-devel

************************************ 
If this email is not intended for you, or you are not responsible for the
delivery of this message to the addressee, please note that this message may
contain ITT Privileged/Proprietary Information.  In such a case, you may not
copy or deliver this message to anyone.  You should destroy this message and
kindly notify the sender by reply email.  Information contained in this
message that does not relate to the business of ITT is neither endorsed by
nor attributable to ITT. 
************************************