Simultaneous Estimation of Attitude and Markov-Modeled Rate Corrections of Gyroless Spacecraft

摘要

The estimation of the rate corrections that are modeled as shortlived Markov processes imparted significant robustness and accuracy to the real-time sequential filter (RTSF) while simultaneously maintaining a small state vector. An extreme example of these qualities is this Kalman filter's ability to converge to the true attitude and rates using only three-axis magnetometer (TAM) data despite starting with no a priori information; this was demonstrated using a wealth of real spacecraft data including some from spacecraft emergencies. The rapid TAM-only convergence of the RTSF at higher angular rates seems to be due to the higher rate of information and may also benefit other algorithms. Last, the rate corrections in the RTSF, although meant to correct dynamically propagated rates without calibrating specific torque or mass models, were nevertheless sometimes useful in identifying errors in control hardware models.