Abstract:
Accurate attitude estimation in high dynamic motion scenarios like unmanned aerial vehicle navigation or human motion and magnetically disturbed indoor environments
requires compensation of accelerometer disturbances as well as magnetometer disturbances. This paper proposes a novel algorithm for accelerometer disturbance rejection by adaptive estimation of accelerometer measurement covariance matrix
online. Magnetic disturbances are generally lasting long compared to accelerometer disturbances which are generally short term in indoor navigation environments. Novel two step Kalman filter update method is proposed to separate attitude correction from heading correction. The magnetometer is used only for heading update such that the attitude estimation is not affected by errors in magnetometer measurement. Performances of the proposed algorithms are shown by using real world sensor data.
