Abstract: In the  process of spacecraft relative navigation, the relative distance measurement information is susceptible to disturbances, which results in greatuncertainty of measurement error. In general, filtering algorithm based on a single model cannot identify the noise and it is difficult to obtain accurate navigation results. As for the limitation of circle orbit  assumption, close relative motion equation (Lawden equation) was studied in the inertial coordinate system using Clohessy-Wiltshire (C-W) equation, and the two equations were established based on the model set. Considering the characteristics of the navigation sensors, a vision based dynamical algorithm of relative position and attitudes estimation using interacting multiple model (IMM-UKF) algorithm was designed based on the Rodrigues parameters and UKF filtering algorithm. The stability and precision of the relative orbit and attitudes determination are certified by using this algorithm under the condition of computation efficiency guarantee.

Key words: Relative position and attitudes, Vision-based navigation, Interacting multiple model, Unscented Kalman filter, Spacecraft