Autonomous orbit and attitude determination for Earth satellites using images of regular-shaped ground objects

Orbit and attitude determination (OAD) is a crucial problem in spacecraft missions. For Earth satellites, especially Earth-observing satellites, images of ground landmarks are feasible to be used for autonomous OAD. Previous studies have demonstrated the possibility of image-based OAD using the light-of-sight (LOS) vector to ground point feature as measurement. However, valid ground point features might not always be available, for example, when the ground landmarks have smooth edges. Besides, when ground landmarks are regular-shaped, this method would discard useful shape information of the objects and affect OAD precision. Aimed at the problem, we present a new OAD scheme using onboard images of regular-shaped ground objects. In the scheme, the LOS vector to the center of the ground objects is used as measurement and is obtained based on the widely-used 3D reconstruction algorithms in machine vision. With the help of an extended Kalman filter (EKF), orbit and attitude parameters can be estimated. The OAD performance of the scheme is assessed using Monte-Carlo simulations. Results demonstrate the feasibility of the scheme and the substantial influence from both lighting constraints and image sampling frequency on the OAD performance. The scheme is also compared with a deterministic scheme, which directly derives position and attitude parameters purely based on the 3D reconstruction algorithm. For the current sensor precision, the proposed scheme is found to have a better performance.