Visual navigation with fast landmark selection based on error analysis for asteroid descent stage

Abstract

Visual navigation is one of the enabling technologies for asteroid landing missions. Quality and efficiency are of paramount importance for determining its robustness and autonomy. This paper proposes a new navigation method, which integrates absolute and relative visual information for quality improvement, and a fast landmark selection method for efficiency purposes. The mapped 3D landmarks with their 2D projections measured in the descent image are primary inputs, and two navigation pipelines, using the extracted 3D-2D and 2D-2D landmark correspondences, are established separately. Then, error analysis, under measurement and system noises, is performed in detail, and based on it, a fast landmark selection algorithm is designed to improve the navigation efficiency. Moreover, the observation model for the attitude motion in adjacent image frames is constructed, followed by the fusion of absolute and relative navigation results using an extended Kalman filter. Finally, Monte Carlo simulations are conducted, and the numerical results demonstrate the advancement of the proposed methods in quality and efficiency.