Abstract:  Autonomous orbit determination,whose goal is to determine the position and velocity based solely on the sensors onboard the spacecraft,is a basic requirement of autonomous operation in formation satellite systems.Relative position measurement is a practical method in autonomous navigation systems while achieving relative position between multiple satellites needs a great number of sensors.For the purpose of satellite payload optimization,a switching measurement scheme was proposed to reduce the number of sensors needed.A general method of autonomous navigation using relative position is the extended Kalman filter (EKF) algorithm.In order to reduce the computational cost of orbit estimators,the filter was subdivided into several parallel sub-KFs,so called multi-stage Kalman filter (MSKF).In addition,in the switching sensor scheme,part of the sub-KFs could be simplified into state prediction only so that the computation load could be further reduced.The number of floating point operations,i.e. FLOPS,was calculated to compare the computational load of MSKF with that of general EKF algorithm.Simulation result shows that the decoupled algorithm has an equivalent performance in navigation accuracy with a much less computational complexity.

Key words: autonomous navigation, switching sensors, multi-stage Kalman filter, decoupled filter, computational performence