Autonomous Lunar Orbit Navigation With Ellipse R-CNN
W.T. Doppenberg (TU Delft - Aerospace Engineering)
Alessandra Menicucci – Mentor (TU Delft - Space Systems Egineering)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
The resurgence of interest in landing on the Moon has sparked the creation of a number of novel technologies concerning Terrain-Relative Navigation (TRN) algorithms. They aid in the need for increasingly precise landing, as well as ensuring fully autonomous operations. To achieve this, most technologies use a ubiquitous feature present on the Moon: impact craters. This research describes the design, development, and testing of an end-to-end TRN system demonstrator that utilises a novel region-based Crater Detection Algorithm (CDA) based on Ellipse R-CNN together with a projective invariant-based crater pattern matching technique that allows for robust ego-position estimation. The system was tested using physically accurate camera views of the Lunar surface. The development of this demonstrator yielded three additions to the field of TRN: a flexible data generation pipeline, a novel AI-based CDA, and results of a novel region-based model in tandem with a modern crater pattern matching technique.