A robust optical navigation method for pin-point landing of asteroid based on 3D navigation features

Visual navigation is an effective way to achieve high accuracy navigation and positioning of asteroids. Selecting effective navigation features and matching them accurately is critical to the visual navigation. The asteroid is small in size, rotates fast, and has a large change in viewpoint between images, thus the image features are significantly distorted. The visual navigation methods based on 2D feature extraction and matching cannot be adapted to such scenarios. Therefore, this paper proposes a visual navigation method based on 3D navigation features. First, the multi-view photometric method is used to produce high-resolution terrain based on low-resolution terrain and high-resolution images taken under different illumination conditions; then, based on a certain reflectance model, the high-resolution terrain is rendered using the current illumination and observation conditions to obtain navigation feature, which is consistent with the texture of the current captured image. Finally, considering the uncertainty of camera pose estimation, a motion-constrained discriminative correlation filter (DCF) is proposed to adaptively match features with distortion. Experiments show that the proposed method can be effectively employed in visual navigation.