Spacecraft Pose Estimation via Monocular Image Processing: Dataset Generation and Validation

Abstract

A new tool capable of rendering high-quality spaceborne synthetic images via ray-tracing is here presented. The assumptions and simplifications used to model both the spacecraft and the background objects are reported, and the trade-off between accuracy and rendering time is discussed, leading to the default settings of the tool. The optical properties of a simplified spacecraft model have been tuned during the qualitative and quantitative validation phase, successfully carried out and detailed in the article. The steps involved in the creation of publicly available labeled spaceborne image datasets with the proposed tool are also provided and discussed in the paper.

Paolo Lunghi

Assistant Professor of Aerospace Systems

Aiming for autonomous Guidance, Navigation, and Control for spacecraft.