This movie presents the performance of an attitude estimation algorithm based on an Extended Kalman Filter (EKF). The EKF algorithm we developed is shown in comparison with the commercial algorithm of the Xsens MTi-G AHRS-IMU.
The test case shown is made with the motors of the quadrotor turned off and after a long while we gave the MTi-G EKF to converge, but with no GPS reception.
Our EKF relies only on the gyroscopes and the accelerometers, neglecting measurements from the magnetometers, GPS and barometer.
both algorithms use the same measurements at real time.
As shown, it is quite easy to make the MTi-G diverge even without turning the motors on. Once the motors are turned on the MTi-G losses it completely and diverges (even without moving it).
@berkandincay
worth the price? depends on your needs, for me the answer will be no. I got similar performence with other standard MEMS which were much cheeper, iNEMO evaluation board for example (by ST).
noammeir 4 months ago
@berkandincay
Hi,
the origin of the data was the sensors in the MTiG for both EKFs. the comparison is between our EKF and the EKF which is built in the MTiG, not between two different sensors. note that the test was carried out without GPS reception. according to Xsense in that scenario the MTi would perform better, I didnt try it.
the sensor itself is a standard MEMS sensor, maybe a bit better noise-wise because of the box wrapping it.
noammeir 4 months ago
Hi nice test, I was thinking to buy this sensor but your video make me doubt. Are both tests were conducted on the same Mti-G sensor and you designed your filter to make it work better. I am going to use it for platform stabilization on a vehichle, is it worth its price??
berkandincay 4 months ago