Accelerometer bias calibration using attitude and angular velocity information

For accurate inertial navigation, an accelerometer should be calibrated before usage to reduce error accumulation. However, if limited or no translational observations are available, which is a technical challenge for many applications such as Mars exploration missions, the accelerometer calibration must be conducted using alternative approaches. This paper proposes a novel accelerometer calibration method using attitude and angular velocity information only. In the algorithm, angular acceleration is measured by a gyro-free inertial navigation scheme to propagate the attitude dynamics and kinematics. The attitude, angular velocity, and accelerometer biases are regarded as states of the calibration system. An extended Kalman filter is employed to estimate the states using attitude and angular velocity outputs from the attitude determination system. A discrete-time model and corresponding covariance matrix are derived. Also, the observability of the calibration system is analyzed. It is shown that the biases are not all observable; only a linear combination of them is observable. Therefore, a recalculation process is introduced to improve the calibration performance. Furthermore, the covariance intersection method is used to determine a consistent estimator. Simulation results demonstrate that the proposed method can effectively calibrate accelerometers with high accuracy and improve the performance of the attitude estimates.