Abstract:

A method for designing accurate autonomous navigation system is proposed in this paper for automated transfer vehicles (ATV). This paper devises an accuracy enhancement approach for traditional integrated navigation of a strapdown inertial navigation system (SINS) and a celestial navigation system (CNS) with the virtual observations. This approach benefits from the use of the fact that ATV is in complete weightlessness in orbit without maneuvers. The improved state equations and virtual observation equations are established based on such a constraint, and an extended Kalman filter is used to accomplish the state estimation from SINS, CNS and virtual observations in different data rates.The simulation results indicate that the proposed algorithm can improve both the position and velocity accuracies without loss of attitude estimation accuracy by about 82.33% and 93.87% compared with standalone inertial navigation, by about 98.35% and 98.72% compared with traditional SINS/CNS integration. The introduction of virtual observations can resist the divergence of position and velocity errors due to inaccurate accelerometer bias estimation effectively. This is of significant importance in engineering because the navigation accuracy is improved without the aid of other sensors, meaning it reduces the reliance on the external information for SINS.

Key words: Strapdown inertial navigation system (SINS), Celestial navigation system (CNS), Integrated navigation, Virtual observation, Automated transfer vehicle (ATV)