SatSplatYOLO: 3D Gaussian Splatting-based Virtual Object Detection Ensembles for Satellite Feature Recognition

TL;DR

This paper introduces SatSplatYOLO, a novel method combining 3D Gaussian splatting and YOLOv5 ensembles to accurately detect satellite components for autonomous space operations.

Contribution

It presents a new pipeline that uses accelerated 3D Gaussian splatting and virtual view rendering to improve satellite component detection in orbit.

Findings

Achieves high-confidence detection of unknown satellite components.

Enables autonomous guidance and control for space missions.

Operates efficiently on-board spacecraft.

Abstract

On-orbit servicing (OOS), inspection of spacecraft, and active debris removal (ADR). Such missions require precise rendezvous and proximity operations in the vicinity of non-cooperative, possibly unknown, resident space objects. Safety concerns with manned missions and lag times with ground-based control necessitate complete autonomy. In this article, we present an approach for mapping geometries and high-confidence detection of components of unknown, non-cooperative satellites on orbit. We implement accelerated 3D Gaussian splatting to learn a 3D representation of the satellite, render virtual views of the target, and ensemble the YOLOv5 object detector over the virtual views, resulting in reliable, accurate, and precise satellite component detections. The full pipeline capable of running on-board and stand to enable downstream machine intelligence tasks necessary for autonomous guidance, navigation, and control tasks.