An All-Parameter System-Level Calibration for Stellar-Inertial Navigation System on Ground

Abstract

The calibration, in particular about the installation errors of a star sensor, is a vital problem when improving the accuracy of stellar-inertial navigation system (INS). This paper proposed an all-parameter calibration method for a ground-based stellar-INS with 12-position rotations, using a Kalman filter that can simultaneously estimate bias, scale factor, misalignments of inertial measurement unit (IMU), and installation errors of star sensor. The difference between the star vector measured by the star sensor and the gold reference generated by the star simulator is used as an observation. On the basis of observability to all parameters, the accuracy is greatly enhanced through an iterative method. Better than previous separate calibration of INS and star sensor, the proposed method offers an advantage in that installation error is slightly influenced by IMU drift and device precision. The experimental results demonstrate that all estimated parameters have good stability and repeatability, with the maximum attitude error of integrated navigation less than ${6^{\prime \prime }}$ after compensation, compared with $2{0^{\prime \prime }}$ using the traditional method. It is shown that the proposed calibration method can efficiently improve the navigation performance of stellar-INS, which has been extensively used in shipborne systems, military aircraft, and missile systems.