Comparison of Spacecraft Vision-Based Navigation Methods and Proposal of a Novel Shadow-Correction Block
A. Bourgeaux and R. M. García Alarcia
Proceedings of the 17th International Conference on Space Operations (SpaceOps 2023), SpaceOps, 2023
Based on my master's thesis (ISAE-SUPAERO), carried out at CNES
Abstract
Since humankind began sending spacecraft to explore our Solar System around 60 years ago, navigation, and operations in general, around and on the explored bodies have evolved substantially. Knowledge on the physics and the environment, and advances in hardware and software systems have been cornerstone to allowing advanced concepts for the execution of interplanetary missions. Indeed, during the last years, new techniques for navigation in such missions have emerged. Among them, Vision-Based Navigation (VBN) has stood out as one of the most promising techniques for navigation around small bodies. This technique leverages on taking images with an on-board camera in the satellite, with the goal of extracting useful information for the relative positioning of the satellite with respect to the navigated body. In general terms, a treatment of the image given by the camera is performed to extract information. Then, this information is transformed into data that can be used by an algorithm of orbit determination, giving the satellite’s state (position and velocity) as a function of time. There are different families of VBN algorithms, some with current in-flight experience (e.g. centroid determination via centre of brightness) and some being proposed in the literature as an enhancement. In this work, VBN algorithms from the family of centroid determination via ellipse fitting, which promises an increase of accuracy over current techniques (at the cost of higher required computational power) have been coded on MATLAB, generating data that is compatible with CNES’ orbit determination tools. The different alternative implementations have been tested and benchmarked with synthetic images of asteroids generated by an internal CNES tool. This work confirms the higher accuracy of this family of algorithms with respect to current centre of brightness methods. It also selects the best algorithmic combination for different scenarios. Additionally, it proposes a novel correction method that diminishes the disturbing effect of large areas of shadow on the navigated body.
Keywords: Vision-Based Navigation, Line-Of-Sight vector, centroid determination, ellipse fitting, small bodies, shadowing correction
Cite
@inproceedings{Bourgeaux2023SpacecraftVisionBased,
author = {Bourgeaux, Aur{\'e}lia and Garc{\'\i}a Alarcia, Ram{\'o}n Mar{\'\i}a},
title = {Comparison of Spacecraft Vision-Based Navigation Methods and Proposal of a Novel Shadow-Correction Block},
booktitle = {Proceedings of the 17th International Conference on Space Operations (SpaceOps 2023)},
address = {Dubai, United Arab Emirates},
publisher = {SpaceOps},
month = {mar},
year = {2023},
doi = {10.82217/spaceops2023_161},
url = {https://publications.spaceops.org/2023/0161},
abstract = {Since humankind began sending spacecraft to explore our Solar System around 60 years ago, navigation, and operations in general, around and on the explored bodies have evolved substantially. Knowledge on the physics and the environment, and advances in hardware and software systems have been cornerstone to allowing advanced concepts for the execution of interplanetary missions. Indeed, during the last years, new techniques for navigation in such missions have emerged. Among them, Vision-Based Navigation (VBN) has stood out as one of the most promising techniques for navigation around small bodies. This technique leverages on taking images with an on-board camera in the satellite, with the goal of extracting useful information for the relative positioning of the satellite with respect to the navigated body. In general terms, a treatment of the image given by the camera is performed to extract information. Then, this information is transformed into data that can be used by an algorithm of orbit determination, giving the satellite's state (position and velocity) as a function of time. There are different families of VBN algorithms, some with current in-flight experience (e.g. centroid determination via centre of brightness) and some being proposed in the literature as an enhancement. In this work, VBN algorithms from the family of centroid determination via ellipse fitting, which promises an increase of accuracy over current techniques (at the cost of higher required computational power) have been coded on MATLAB, generating data that is compatible with CNES' orbit determination tools. The different alternative implementations have been tested and benchmarked with synthetic images of asteroids generated by an internal CNES tool. This work confirms the higher accuracy of this family of algorithms with respect to current centre of brightness methods. It also selects the best algorithmic combination for different scenarios. Additionally, it proposes a novel correction method that diminishes the disturbing effect of large areas of shadow on the navigated body.},
keywords = {Vision-Based Navigation, Line-Of-Sight vector, centroid determination, ellipse fitting, small bodies, shadowing correction},
file = {2023/2023_Bourgeaux_SpacecraftVisionBasedNavigationShadowCorrection.pdf}
}