Visual Navigation and Shape Reconstruction for Asteroid Approaching Phase

Visual navigation and shape reconstruction are of paramount importance for the safety and success of asteroid missions. This paper proposes a new method using silhouette and pixel area measurements to simultaneously estimate the position of probe and the shape of target during approaching an asteroid. First, the silhouette of target projected on the image plane is extracted using segmentation algorithms, and its pixel area is also calculated for subsequent processing. Second, the relative position of the probe concerning the target asteroid is normalized and estimated based on the pixel area on the hypothesis of uniform linear motion. Then, a parametric shape model of the target is established with a one-dimensional vector and is estimated and fused by minimizing the reprojection error of the silhouette. Finally, synthetic image sequences are generated and Monte Carlo simulations are conducted to verify the proposed method, and the numerical results demonstrate its advances in accuracy.