Yaw Drift with U71D

[Sonia]

We were using an older batch of UM7’s with firmware version U71C. With the magnetometer turned off, after zeroing the gyro, we were pretty happy with the yaw drift (0.1-1 degree per minute).

We’ve just started playing with a newer batch of UM7’s with firmware version U17D. With the same configuration, the yaw drift was a lot higher (2-5 degrees per minute). Even when I run the calibration following the instruction here: http://www.chrobotics.com/forums/topic/continuous-drift-in-yaw-euler, the yaw continues to drift a lot.

Only when I downgrade the firmware version to U71C do I see the small drift again. I’m fine with downgrading, but just wondering if anyone else has noticed this, and if I’m just doing something fundamentally wrong, and if there are more optimal solutions.

Thanks all!

[Imen]

Your problem in the variance setting value. The new firmware version U71D variance value (0.01 gyro, 0.01 acc, 0.005 Mag) are not appropriate to you conditions. If you increase the magnetometer variance (0.01 gyro, 0.01 acc,100 Mag) you will thrust more the magnetometer in the Yaw measurements. You will acquire stable yaw angle with no drift. But this configuration is very sensitive to magnetic Flux distortion. You can overcome the magnetic flux distortion by Mag calibration. In fact, I use the following value of variance in my application and it work with no drift. (0.01 gyro, 0.01 acc,100 Mag). I repeat the Mag calibration for any problem through:
1. Magnetic calibration using the CHR Serial Interface tool
2. Zero the gyros (CHR Serial Tool command)
3. Set the magnetic reference (command). Point your UM7 to the North before setting the magnetic reference
4. Reset the EKF (command).

It shoud work.
feel free to ask any other question

[Author]

Posts