Centroiding Error Compensation of Star Sensor Images for Hypersonic Navigation Based on Angular Distance

A celestial navigation system (CNS) can be integrated in vehicles to suppress the diverging error of the inertial navigation system (INS). However, the CNS is subjected to aero-thermo-optic effects in hypersonic environments that introduce large-scale centroiding errors in star sensor images and greatly reduce the star sample size, limiting the usefulness of the CNS for hypersonic vehicles. In this study, these difficulties were addressed by the development of a statistical model for measurement noise based on the angular distance so that the standard deviation (SD) of the measurement noise can be correlated with the star sample size. To compensate for the centroiding error, a variable-dimension pseudo-transient adaptive filter was designed where the initial R matrix is set according to the star sample size. A semi-physical experiment was conducted to evaluate the performance of the proposed method, and the results indicated that stable centroiding error compensation was achieved for extremely small star sample sizes. In particular, the proposed method compensated for up to 66.24% of random positioning error, which is a performance improvement of about 10% compared with conventional compensation methods.