||Charles County RC Field
||Rev 2.2 with CVS head as of 2002-09-22
||Flyer and Attitude from CVS head
Joe Easly made movies! They're Quicktime, so you'll need the new mplayer
to view under Linux. I'm not sure which flights these correspond to,
but they're fun to watch. Without Joe telling us when the autopilot
is engaged, it is hard to tell the difference. It really is working
This is the second day of autonomous flight with the flyer program
doing attitude hold and attitude command mode. We had eight flights,
with over twenty minutes of autonomous hovering.
I haven't had time to produce good graphs from each flights since the
output format has changed from the previous flights.
Pitch attitude control tests showed that the 2 deg limits on the
controller were too small. The controller hit its "stops" while
trying to maintain its attitude.
We increased the pitch control limits to 3.5 degrees and the roll limits
to 4 degrees. We also fixed the sign on the roll control feedback.
Aaron noticed the sign on the roll feedback was backwards, so we
fixed it. The flight ended with a gust of wind sending us backwards
too quickly, so Aaron took back manual control.
Roll tests on this flight were fine and showed no significant divergence.
It hovered for two minutes through significant wind gusts with no
problems. We eventually got bored and had to refuel.
We enabled both pitch and roll for this flight. It did a great job,
but quickly maxed out the controller's limits due to the wind.
Aaron couldn't get enough attitude deflection to overcome it
quickly, so the helicopter sort of wallowed about.
After three minutes of hovering, we let it try a landing.
Aaron eased the collective down to about six inches, then dropped
it to plant the skids. The roll controller tried to maintain
the hoveirng, skid-low attitude and maxed out its limits.
We'll need a squat switch or something to disconnect the autopilot
We increased both pitch and roll authority limits to 5 degrees.
This helped out quite a bit. We did a 2 minute hover that drifted
aways, then flew back manually. Aaron re-engaged the autopilot
and let it hover for another two minutes. He did another auto
We increased the attitude command authority for Aaron from 5 to
+/- 8 degrees. This produced more snappy controls, rather than the
slower response that the lower settings allowed. We then flew an
entire tank of fuel all the way to an automatic landing.
This time we tried to add a small amount of integrator state to the
attitude controller. The roll axis PID loop was set to I=0.1 and
the pitch I=0.06. Aaron let it hover for ten seconds, but the
integrators added quite a bit of instability. He brought it
down and we disabled the integrators.
With the integrators off, Aaron took off and had the aircraft do some
high speed maneuvers. He also tried giving it impulses to see how well
it could recover. The controllers recovered well for forward pitch and
left roll, but backwards pitch and right roll did not recover quickly.
After burning a full tank of fuel, Aaron let it autoland and we called
it a good day.