Posted by Edouard H on November 12, 2009 at 2:36pm
Hi all,I am wondering if it would be possible to use a magnetometer-only solution to estimate UAV attitude. Indeed, it looks like there are quite good sensors available e.g micromag 3 which would allow to estimate plane attitudes quite accurately and at a reasonable frequency.Given the range of our UAVs, magnetic declination can be considered as fixed.I am worried however about the capability to compensate magnetic perturbations generated by the engine and electronic equipments on the plane. Compensation for these perturbation would most likely be required. Compensation would also probably need to be adjusted with the throttle setting...Did anybody in the community experiment with this? I would be keen to get some input before i go and try this out myself.
You need to be a member of diydrones to add comments!
If you're on a very tight budget, and all you want to experiment with is a camera stabilization system, you may want to take a look at this page. It (the stabilized servo) may be a lot less expensive than hobbling parts together on your own (although probably not as much fun)!
Interesting idea :-)
The core answer seems to be that while you can know what direction you're pointing, you can't know what direction "down" is, and therefore can't know whether a control movement will send the thing up, left, right, or down.
I'm fairly new to this, but if your issue is just camera stabilization on a pan/tilt, I'd guess a 2-axis gyro would be the minimum and possibly all you need.
Will,
You are right indeed. When the plane is aligned with the Earth magnetic field, it seems pretty obvious you won't been able to estimate roll...
Having realised this, I believe i could still use the GPS velocity to complement the Magnetic field reference and still manage to get attitudes w/o Accels and Gyros. If it works, it could be interesting as a poor man's IMU.
Dan,
I am very interested by your calibration trials. Do you ave an OOM of the perturbations you measured compared to earth Magnetic field? Did you try positioning teh Micromag away from the rest (tail)?
As Bill pointed out, the single reference vector will not do the job. A 3-axis magnetometer will define a 3D magnetic vector but any aircraft rotations about this vector will be unknown.
I have experimented with a MicroMag3 mounted amongst other electronics in an Easy Star. By recording 3 axis data as you randomly move the aircraft through different orientations, you can plot the resulting sphere and compare it to the expected sphere. At 0% throttle, the sphere was offset slightly, whereas full throttle both offset and distorted the sphere. I was able to calibrate out these distortions with respect to throttle using a linear approximation but I never thoroughly tested the performance.
Replies
A lightweight low-power magnetometer only based UAV flight control system
Interview with Matt Chave by Chris Anderson Part 1:
http://diydrones.com/profiles/blog/show?id=705844%3ABlogPost%3A22568
Interview with Matt Chave by Chris Anderson Part 2:
http://diydrones.com/profiles/blogs/705844:BlogPost:22575
Matt's paper describing Apeliotes - Magnetometer based UAV flight control system and it's issues:
http://www.archive.org/download/A_lightweight_lowpower_magnetometer...
The core answer seems to be that while you can know what direction you're pointing, you can't know what direction "down" is, and therefore can't know whether a control movement will send the thing up, left, right, or down.
I'm fairly new to this, but if your issue is just camera stabilization on a pan/tilt, I'd guess a 2-axis gyro would be the minimum and possibly all you need.
You are right indeed. When the plane is aligned with the Earth magnetic field, it seems pretty obvious you won't been able to estimate roll...
Having realised this, I believe i could still use the GPS velocity to complement the Magnetic field reference and still manage to get attitudes w/o Accels and Gyros. If it works, it could be interesting as a poor man's IMU.
Dan,
I am very interested by your calibration trials. Do you ave an OOM of the perturbations you measured compared to earth Magnetic field? Did you try positioning teh Micromag away from the rest (tail)?
As Bill pointed out, the single reference vector will not do the job. A 3-axis magnetometer will define a 3D magnetic vector but any aircraft rotations about this vector will be unknown.
I have experimented with a MicroMag3 mounted amongst other electronics in an Easy Star. By recording 3 axis data as you randomly move the aircraft through different orientations, you can plot the resulting sphere and compare it to the expected sphere. At 0% throttle, the sphere was offset slightly, whereas full throttle both offset and distorted the sphere. I was able to calibrate out these distortions with respect to throttle using a linear approximation but I never thoroughly tested the performance.
Dan
You need at least two non-colinear reference vectors to establish attitude, so a magnetometer by itself will not do the job.
What is sometimes used for helicopter applications is a combination of a 3 axis gyro, 3 axis accelerometer, and 3 axis magnetometer.
For a fixed wing aircraft, you can use a 3 axis gyro, 3 axis accelerometer, and a GPS.
You might want to read Robert Mahoney's papers, or a summary of the key concepts.
Bill