Autonomous celestial orbit determination using Bayesian bootstrap filtering and EKF

The celestial navigation system is one of the important autonomous navigation system for spacecraft. The basic principle of it is using the extended Kalman filter (EKF) and the measurement of angle between celestial bodies estimate the position and velocity of spacecraft. But the EKF may not converge because of the inaccurate initial state. In this paper, a new Bayesian bootstrap filtering approach is used at the initial stage of the filtering interval to provide EKF with an accurate initial state to overcome the divergence problem. Simulation results demonstrated the high precision of orbit determination even in the presence of large initial states errors.