Imaging-based attitude determination algorithm for small satellites: Design and the preliminary results

Güzel M. B. , Söken H. E. , Tekinalp O.

71st International Astronautical Congress, IAC 2020, Virtual, Online, 12 - 14 October 2020, vol.2020-October identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 2020-October
  • City: Virtual, Online
  • Keywords: Camera, Earth images, Imaging-based attitude estimation, MEKF, Small satellites


© 2020 by the International Astronautical Federation (IAF). All rights reserved.Limitations for small satellite missions encourages researchers and engineers to use on-board equipment for multiple tasks. Thus, both space can be saved and the equipment redundancy can be prevented. For Earth-imaging satellites a prominent equipment that can be used for multiple purposes is the camera. Other than the observation, the taken images can be used as measurements for determining the absolute or relative attitude of the spacecraft. Depending on also the content of the image (e.g. stars, Earth, Moon) a camera can be an alternate attitude sensor that is capable of providing high accuracy attitude measurements in low cost. This research aims at designing an attitude estimation algorithm, which can aid the coarse attitude estimates with the attitude information extracted from Earth images. Whenever an image is available to be used for attitude estimation, reference Earth data is matched with the ellipsoid that Earth image forms on the sensor frame to estimate the three-axis attitude of the spacecraft. This estimated attitude is fed to a multiplicative extended Kalman filter (MEKF) algorithm that runs as the main block of the attitude estimator. The measurements from the conventional attitude sensor block, which is formed of a three-axis magnetometer and Sun sensor, are pre-processed with QUEST algorithm to unify the measurement update process of the filter. This paper introduces the architecture of the proposed algorithm together with the details on possible methods to extract the attitude information from the obtained Earth images. Preliminary results to demonstrate the success of the algorithm are presented.