Two-stage framework for ultra-close-range spacecraft pose measurement

To address the urgent need for ultra-close-range spacecraft pose measurement in space missions, such as rendezvous, docking, and capture, this paper proposes a two-stage pose measurement framework fusing target detection and spatial circle pose optimization, effectively tackling the challenges of pose measurement under complex lighting conditions at ultra-close range. First, an improved target detection algorithm is employed to achieve rapid and robust localization of cooperative target regions across several typical lighting conditions. Subsequently, leveraging the spatial conical geometry model of camera imaging, a nonlinear optimization model based on spatial circle reprojection error is constructed to iteratively refine the initial pose estimates, enhancing measurement accuracy even under challenging lighting conditions. Experiments demonstrate that the improved target detection algorithm reduces the parameter count by 47% compared to YOLOv8s while maintaining detection accuracy; our pose optimization method improves attitude estimation accuracy, enabling real-time position measurement errors <4.22mm and angular errors <0.58^∘ for ultra-close-range spacecraft, verifying the framework's fast and high-precision characteristics. This study provides a feasible visual technical solution for ultra-close-range spacecraft pose measurement in complex environments.