Reliable Onboard 3D Reconstruction of Unknown Spacecraft via Data Acquisition Guided by Orbital Geometry and Lighting

We present a framework for autonomous 3D reconstruction of non-cooperative spacecraft in Low Earth Orbit (LEO) using a single chaser satellite performing impulsive-burn inspection maneuvers. Rather than relying on swarms, formation flying, or predetermined waypoint paths, our approach focuses on strengthening two key components: (1) improving the reliability of Structure-from-Motion (SfM via COLMAP) under challenging orbital lighting conditions; and (2) characterizing data acquisition parameters that directly affect reconstruction quality. SfM serves as a prerequisite for downstream 3D Gaussian Splatting (3DGS) and is made more robust through targeted algorithmic tuning and removal of imagery degraded by eclipse transitions. Using a high-fidelity digital twin environment incorporating Hapke surface response and ephemeris-driven lighting, we generate time-accurate inspection sequences and evaluate how image resolution and frame sampling influence SfM completeness and 3DGS performance. This end-to-end study provides guidance for selecting onboard image acquisition settings and demonstrates improved reconstruction stability across realistic LEO inspection conditions. Performance is validated using implementations compatible with onboard execution for autonomous inspection and reconstruction.