Visual Navigation Algorithm of Small Body Lander Based on UFIR Filter

Abstract

The gravity of small body is small and its distribution is not uniform. Therefore, for small body lander, the process noise in traditional Kalman filters is difficult to count. In addition, due to the complex environment of deep space, there is no guarantee that the visual navigation measurement noise must be Gaussian white noise, and it is difficult to calculate statistical knowledge such as its covariance. These problems will affect the estimation accuracy of the Kalman filter. In this regard, this paper combines a crater-based navigation algorithm, uses the UFIR filter that does not require process and measurement noise statistics. Colored noise is added to process equation and measurement equation for simulation, and compare the estimation errors of Kalman filter and UFIR filter. The results show that when the process noise and measurement noise are non-Gaussian white noise or the prior statistical knowledge of noise statistics is inaccurate. The UFIR filter has better estimation effect and is more robust.