This study developed a real-time spacecraft pose estimation algorithm that combined a deep learning model and the least-squares method. Pose estimation in space is crucial for automatic rendezvous docking and inter-spacecraft communication. Owing to the difficulty in training deep learning models in space, we showed that actual experimental results could be predicted through software simulations on the ground. We integrated deep learning with nonlinear least squares (NLS) to predict the pose from a single spacecraft image in real time. We constructed a virtual environment capable of mass-producing synthetic images to train a deep learning model. This study proposed a method for training a deep learning model using pure synthetic images. Further, a visual-based real-time estimation system suitable for use in a flight testbed was constructed. Consequently, it was verified that the hardware experimental results could be predicted from software simulations with the same environment and relative distance. This study showed that a deep learning model trained using only synthetic images can be sufficiently applied to real images. Thus, this study proposed a real-time pose estimation software for automatic docking and demonstrated that the method constructed with only synthetic data was applicable in space.

CubeSat programs are often conducted under severe constraints in budget, personnel, schedule, and knowledge continuity, while still requiring sufficient verification to support mission success. This paper presents the verification, testing, on-orbit operation, and anomaly analysis of COSMIC, a 3U CubeSat jointly developed by Yonsei University, Unmanned Exploration Laboratory (UEL), and UZURO Tech Inc. in less than eleven months from project kickoff to launch. COSMIC adopted a protoflight model approach, prioritized commercial off-the-shelf components with flight heritage and prior laboratory experience, and concentrated limited resources on subsystem, FlatSat, system-level, environmental, communication, and scenario-based testing. The test program is reviewed against the subsequent on-orbit results to identify which verification activities contributed to anomaly prevention, fault isolation, and recovery. The results show that system-level integration testing, thermal vacuum cycle testing, and scenario testing were particularly effective in revealing interface, software, and deployment-related issues before launch and in supporting early orbit recovery. However, the mission also exposed limitations in recovery-logic verification, long-duration software testing, communication robustness, and independent reset paths. Based on these findings, practical recommendations are derived for resource-constrained CubeSat teams seeking to tailor verification activities without simply reducing test rigor.

The authors have made the following corrections to the author list and the content of the original publication