Feature Collaborative Derivation Method for Small Celestial Body Landing

Autonomous navigation is the key technology for the success of the small celestial body landing mission. The significant terrain features on the surface of small celestial body can provide position and attitude information for autonomous navigation. In most existing optical navigation methods, a sufficient number of features are required to guarantee that the lander's state is observable. However, the landing area is usually flatter, which results in fewer significant terrain features that the camera can observe. Meanwhile, in the process of small celestial body landing, the observation range of the camera will be reduced with the decrease of the height, and the information that can be observed by a single camera is also decreased. To address this problem, a feature collaborative derivation method with limited observation is proposed in this paper. In this method, the sequence images of a single camera are used to predict the vanished features. Meanwhile, the overlapped observation areas of multiple cameras are matched to realize the collaborative observation of features. Then, the time sequence prediction value is fused with the collaborative observation value, so as to obtain the collaborative derived information of vanished features with higher accuracy. The effectiveness of the proposed feature collaborative derivation method is verified through numerical simulations in a small celestial body navigation scenario.