Non-traditional robust UKF against attitude sensors faults

Söken, Halil Ersin
Hacızade, Cengiz
Çilden Güler, Demet
Two non-traditional approaches for nanosatellite attitude estimation are investigated. In the non-traditional approach for attitude filtering the magnetometer and Sun sensor measurements are first used in the Singular Value Decomposition (SVD) algorithm to determine the spacecraft's attitude. Then the attitude estimates from the SVD are fed into the Unscented Kalman Filter (UKF) for getting finer attitude estimates and estimating the attitude rate. In the first case, this algorithm is used without any further modification for sensor faults. In the second case the SVD-aided UKF is adapted against the sensor measurement faults. For adaptation the filter's measurement noise covariance (R) matrix is tuned with a scale factor. The filters-the SVD-aided UKF (SVD/UKF) and the SVD-aided Robust UKF (SVD/RUKF)-are tested in two different fault scenarios: constant continuous bias and measurement noise increment. Fault scenarios are repeated for both sun sensor and magnetometer. The results indicate the non-traditional approach is inherently robust against the sensor faults as vector measurements are first pre-processed in the SVD before running the UKF.
Citation Formats
H. E. Söken, C. Hacızade, and D. Çilden Güler, “Non-traditional robust UKF against attitude sensors faults,” Matsuyama, Japonya, 2017, p. 1, Accessed: 00, 2021. [Online]. Available: