Need advide RE attitude measurement

[chrisgornall01]

Hi, I'm an electronic engineer who's a total newbie with regards to UAVs.
I am starting to develop a system and have a few questions regarding the best sensors to be using and their implementation.

My plan is to use 2 accelerometers to measure pitch and roll, and a mercury switch to detect an inverted state. The problem I can see is one of the sensor readings being incorrect during an attitude change due to the momentum of the aircraft and due to turbulance. Is this actually a factor worth worrying about?
Which sensors do you all find best for the task?

I'm not planning on autonomous landings, and am debating the necessity of yaw sensing and correction, what are peoples opinions on this?

I'm also guessing that, on the matter of heading, this is best done by extrapolating from previous GPS co-ordinates. Is this correct, or is some form of electronic compass a worthwhile addition.
I'm going to be taking my time on this project and putting considerable effort into it, so I don't want to be cutting corners unnecessarily or neglecting useful features.

Thanks in advance for any help, Chris.

:) :) :)

/Edit Processing power is not going to be a major factor as there will be plenty of computational power onboard, and anything with a high overhead will be passed off to co-processor boards.

[alexcmag]

Chris,

I'm working with some friends on a simple autopilot project ( http://www.rcgroups.com/forums/showthread.php?t=673948 ) and accelerometers were suggested too.

I suppose you're thinking on using accelerometer as a level meter (i.e., to check where the gravity force is). This doesn't work with airplanes.

On a coordinated turn an airplane (and everything inside it) will sense the gravity as it is on horizontal. That's why passengers can enjoy their drinks without trouble even when airplane is on a turn.

Watch this video and you will understand: http://www.youtube.com/watch?v=Xp2Uc9XvmjY

Most RC stabilization systems are based on visible light or infrared sensors to "see" where is the sky and where is earth, but of course it will not work well near mountains.

In our case we're planning to start with aerodynamic stabilization (a well designed airplane with enough dihedral and aft CG).

I'm also searching for effective ways to measure pitch and roll. Analog Devices has a set of inertial sensors that can help a little: http://www.analog.com/en/prod/0%2C2877%2CADIS16350%2C00.html , but they probably will not give good readings over the time. Maybe used with simple accelerometers to check gravity vector can produce good results.

[chrisgornall01]

Hmm, I'd like for it to work in low light conditions, horizon sensing will be pretty ineffective then...
How is it done in the commercial ones?

[joelhaasnoot]

Gyroscope + Accelerometer together at a minimum, in combination with GPS and some have magnetic compasses... (check out the links section under 'Components' and 'Autopilots')

[Gary Warner]

Hi Chris,

Accelerometers, like said above, can relay 'perceived' gravity. The polar displacement (combined result of forward, side and vertical displacement relative to a fixed point inside the plane) and amount of the gravity seen (amount of displacement) by the accelerometers can be useful in flight assistance or UAVs'. I use accelerometers to establish the function of a slip/skid indicator (inclinometer) and use this to control yaw. So far as pitch and roll, gyros seem to be the primary sensors, backed up with accelerometers. The results of using accelerometers alone to control the pitch and roll stuff is well demonstrated in the above video and would not be what you are trying to accomplish. Plainly put, most stable systems for flight assistance or UAV's involves the combined use of both the accelerometer and the gyro.

Oh, and the mercury switch can be substituted with the veritcal axis reading for an accelerometer and used in code to perform what ever function you want when negative G's are seen.

[chrisgornall01]

I see, sorry to keep bombarding you all with questions, but how do the available gyros perform with regards to drift? I really want to avoid using horizon sensing if at all possible.

[alexcmag]

Analog Devices has a set of gyros and accelerometers you can use.

I liked the Gary's approach. Solid state gyros will not have precision enough to track attitude for a long time, but when they show you're not having any pitch, yaw or roll movement this means you probably can thrust the accelerometer to "fix" the position.

AD gyros:

http://www.analog.com/en/subCat/0,2879,764%255F801%255F0%255F%255F0%255F,00.html

AD accelerometers:

http://www.analog.com/en/subCat/0,2879,764%255F800%255F0%255F%255F0%255F,00.html

[chrisgornall01]

Thats what i was presuming, as the UAV will be tracking waypoints, most of the time should (hopefully) be spent in level flight.
When the gyros are showing level flight within tolerances, the accelerometer output could possibly be used to calibrate the gyros.

If we could come up with a unit which provides absolute measurement of pitch and roll, then this hobby would be so much easier and cheaper for the average person to start out in.

[4mla1fn]

just found this thread. i'm in the same boat as the OP, except i didn't have his foresight. just got my proto board (gps, 3axis mag, 3axis accel, baro, temp) working and i realized that the accel alone will give garbage on a plane. time for some gyros, i guess. man, those suckers are expensive. :(

[chrisgornall01]

I'm looking at the IDG300, http://www.sparkfun.com/commerce/categories.php?cPath=23_85
Seems like a reasonable price.
Think we can get everything we need using this, 3 accelerometers, and 1 magnetic compass.

[joelhaasnoot]

You'd only need 2 of the IDG300 sensors - they're 2 axis (assuming you want to track all 3 axis), and Sparkfun also has a three-axis accelerometer chip - meaning you'd only need one of those...