> One other thing I forgot to mention is that I'm also considering
 > an R/C
 > heading-lock gyro. I'm not sure exactly how these work, however... Any
 > experience, anyone?


Don't forget your biggest problem -- you have to account for angular
momentum (a.k.a. centrifugal force) in a turning aircraft.

Remember in a fixed-wing aircraft in a coordinated (important... are all
turns your UAV makes coordinated with correct aileron and rudder
inputs?) 60-degree bank turn, an accelerometer will register an almost
perfect 2G constant "pull" directly toward the floor of the aircraft
which is of course, 60 degrees off of level.

If the aircraft is banked very slowly and in a coordinated fashion,
"occupants" like the accelerometer will barely "feel" the entrance to
the bank.  This is one of the human-factors causes of the classic
high-speed spiral dive when operating in IMC without a horizon
reference.  Thus you need another input which can sense this banking
tendency.

I think you're going to have to use a combination of techniques to sense
"level" in an aircraft of any type.  The standard solution is to use a
gyroscope which is leveled on the ground prior to aircraft motion, of
course.  The smaller the gyro and the more internal friction in the
bearings, the more it will tend to "precess" or tip and stay tipped
after aggressive maneuvers.

High-dollar UAV's (i.e. military) use a variety of inputs from redundant
gyros (usually ring-laser type - light doesn't have friction so
precession errors are nil), GPS, and other flight data sensors
(airspeed, compass, standard navaids) to determine the exact attitude of
the aircraft in relation to the ground and to ignore inputs from sensors
which have failed or are sending data that "doesn't make sense" to the
flight management computer(s).  The really high-end stuff can take these
inputs and calculate aircraft position and orientation even if the
system reset in flight by doing "odd-man-out" anaylsis of the data.

A much more practical way to keep a UAV upright in a tight budget would
be to design a wing with a very high degree of dihedral -- many training
RC aircraft used to have this, thus giving the aircraft a tendency to
right itself.  Of course, too much of this tendency toward stability
will make the aircraft impossible to bank if not enough aileron and
rudder force is applied... and applying more aileron requires more
rudder input for a coordinated turn, so you end up at a mathematical
slippery slope where you can overdesign the stability of the wing in
relation to the force you have available from any of the movable control
surfaces.  Catch-22.

Also depending on whether or not your aircraft will have any tendency to
leave controlled flight (stall/spin, etc.) you may find that some gyros
are going to reach a mechanical lockout point where they "tumble" as
they reach the end of their mechanical travel.  If your software cannot
account for a gyro failure (loss of electricity for powering the gyro, a
tumble, etc.) it may send your aircraft wildly out of control when the
gyro fails.

I would say perhaps some of the gyroscopic stabilization systems for RC
hellicopters might be appropriate for a relatively simple commercial
solution to your problem, but they will have published limits.  I've
read about such systems but never used one on an aircraft.

The pendulum won't work for the reason mentioned above about angular
momentum.  In that same 60-degree bank turn, the pendulum for all
intents and purposes will be hanging pointed directly toward the "floor"
of the aircraft.

Heading lock gyro's are simply turning in a vertical plane instead of a
horizontal one, and typically they have a 360 degree rotation
capability, continuously.  (i.e. the aircraft can continue to turn
around and around and around and you won't lock the gyro.)  Heavy pitch
changes in the horizontal plane will cause friction at the ends of the
gyro, and will cause this gyro to precess.  A "heading hold" type of
gyro doesn't care much about precession since you don't reference it
until the aircraft is in stable, relatively level, flight.  Then you
have the aircraft bank appropriately to return the gyro to the original
position.  Over time, that heading will wander one direction or the
other, depending on where the friction is in the gyro system.

It will be interesting if you continue with the accelerometer idea as to
how you overcome the angular momentum issue.  That comment about using
the IR detector to tell where "up" and "down" are is interesting in a
low-cost solution.  It will be fun to see what you come up with.

--
Nate Duehr, nate@natetech.com

Quando Omni Flunkus Moritati
"When all else fails, play dead."

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu