This paper proposes a method for self-calibrating the star sensor on the Mars rover considering several years of exploration on the surface of Mars. The natural stars in the night sky are considered the control points, and a self-calibration model is deduced in detail according to an imaging model. An adaptive celestial positioning (ACP) algorithm is then introduced, and the calculation procedure is presented in detail to realise self-adjustment based on the self-calibration of the star sensor. Three field tests were conducted on Earth, the results of which show good self-calibration and celestial positioning performances. The positioning results indicate an obvious accuracy improvement using the ACP algorithm compared with that without calibration. Multiple positionings in one night can improve the celestial positioning accuracy to approximately 15 m. For future studies, this self-calibration model will be useful not only for star sensors but also for other optical sensors, such as sun sensors and binocular or stereo-vision cameras.

A new integration of the acquisition and tracking modes is proposed for the integration of a Celestial Navigation System (CNS) and a Strapdown Inertial Navigation System (SINS). After the integration converges in the acquisition mode, it switches to the tracking mode. In the tracking mode, star pattern recognition is unnecessary and the integration is implemented in a cascaded filter scheme. A pre-filter is designed for each identified star and the output of the pre-filter is fused with the attitude of the SINS in the cascaded navigation filter. Both the pre-filter and the navigation filter are designed in detail. The measurements of the pre-filter are the positions on the image plane of one identified star. Both the starlight direction and its error are estimated in the pre-filter. The estimated starlight directions of all identified stars are the measurements of the navigation filter. The simulation results show that both the reliability and accuracy of the integration are improved and the integration is effective when only one star is identified in a period.

A strapdown inertial navigation system and celestial navigation system integrated autonomous navigation scheme is proposed in this paper, using the navigation information obtained from Earth sensors and star sensors. To eliminate the adverse effect caused by the asymmetry of Earth infrared radiance, the relationship between Earth infrared radiance brightness and effective horizon height is found. According to the relationship as well as the measuring principle of the Earth sensor, this paper derives a function to correct the measurement of the Earth sensor. Then, the angle-distance of stars can be calculated, and using this information, we can estimate the navigation information of a ballistic missile by least square estimation. The simulation results show that the error of Earth infrared radiance has a great effect on the navigation precision, and by using the correction scheme, this adverse effect can be greatly mitigated. This correction scheme is available and effective.

A guide star catalogue characterised by fewer guide stars, uniform distribution and high completeness is conducive to the improvement of star pattern identification and star tracking efficiency, and it has become an important study objective. A screening method, which takes “quasi-uniform distribution” of space solid angles as the principle and the size of the space solid angle corresponding to 4°×4° on the equator as the reference and divided the whole celestial sphere into 2,664 sub-blocks in sequence, is proposed in this paper. Based on this method, a “quasi-uniform” guide star catalogue with 2,937 guide stars was obtained. According to our ergodic statistical analysis of the whole celestial sphere, in a 12°×12° field of view, after the space solid angle method was used to divide the celestial sphere, the probability of emergence of three guide stars was 99·9% or above, while the number of guide stars decreased by 12·6% compared to the inscribed cube method. It can be concluded that when the completeness is equal, the space solid angle method is superior to the inscribed cube method in both capacity and distribution uniformity.

The accuracy of attitude determination using a star sensor tends to be degraded if the host vehicle's manoeuvring smears the star image. In this paper, a new restoration algorithm with the aid of a Strap-down Inertial Navigation System (SINS) is proposed to reduce the effect of the smeared image. The smeared trace length is estimated with aid of the SINS angular rate. The restoration algorithm based on a Wiener filter is designed after the smeared zone is derived with the aid of the SINS coarse attitude. A tracking method is proposed to reject the stars of low centroid extraction accuracy. Simulations demonstrate that the success rate and the accuracy of attitude determination are improved significantly by the restoration algorithm even under a very fast rotation. The impact of the SINS error on the restoration is evaluated in the simulations.