On Tue, Aug 26, 2008 at 7:02 AM, Csaba Halász wrote:
> On Tue, Aug 26, 2008 at 6:08 AM, Gonzalo Rubio wrote:
> >
> > The objetive of the project is to
> > fly both the real plane and the simulator at the same time and together
> with
> > and adquisition data system compare both reality and virtual performance.
>
> I believe Curt does similar things.
I've done something similar, but only with a pretty crude level of detail.
Still, I'm fairly proud of the results.
The first UAV platform I worked with was a Sig Rascal 110 (110" wing span.)
With the help of aeromatic and one of the JSBSim developers (David?) we
developed an initial approximate JSBsim model for the Rascal, and then based
on my real life flight experience, I refined some of the model
characteristics to get more realistic adverse yaw, roll rates, more
realistic pitch moment relative to throttle change, etc.
I then used this simulator model to develop a set of autopilot
configurations (stages, gains, throw limits, etc.).
At the same time, I was developing a flight computer that used the
flightgear autopilot code (and property system, and xml parser.) By the
time the flight computer was ready for real flight tests, I had moved on to
an electric powered senior telemaster (8' wing span) for my development
platform. The two airframes were similar enough, that the gains and setup
developed with the Rascal simulator model worked out of the box on the real
telemaster. I had stable wing leveling and stable pitch hold on my very
first real world attempt with the Senior Telemaster. I was very proud of
that result. The dynamics of the rascal and telemaster are different
enough, that I had to do a bunch of subsequent tuning to optimize my
controller, but the fact that it was convergent on the first try was really
cool. (Worst case scenario is that the controller diverges and you begin
shedding airframe parts in flight.) :-)
For the future, I want to experiment with incorporating the nasal script
engine into my flight computer. With that in place, I want to be able to
write scripts to auto fly various flight test scenarios. The goal here
would be to develop a set of very accurate and repeatable flight test
profiles that are flown by the flight computer (as directed by the script
engine at a high level). I think this would be a very useful and very
powerful for characterizing the dynamics of a UAV. There's a chicken/egg
issue here, but if you develop and refine the simulation model in parallel
to developing and refining your autopilot configuration, I think you could
boot strap yourself pretty quickly. And if we start building up a set of
proven autopilot configurations for a variety of platforms, there will be an
increasing likelihood that just about anyone can find a working start
configuration, and then be able to refine it from there.
Regards,
Curt.
--
Curtis Olson: http://baron.flightgear.org/~curt/
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Flightgear-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/flightgear-devel
