@phdthesis{oai:kyutech.repo.nii.ac.jp:00006344,
 author = {Amartuvshin,  Dagvasumberel},
 month = {2020-01-22},
 note = {1 Introduction||2 Background||3 Earth-horizon based Approach||4 Visual-Inertial Attitude Propagation Approach||5 Map-based Approach||6 Conclusions, Attitude determination, along with attitude control, is a critical component for many Earth-observation small satellites. However, most of the small satellites operate without redundant attitude sensors due to the satellite’s small form factor, so that there is a significant risk of mission failure. Moreover, the range of possible applications for small satellites is limited due to their low available volume, weight, power, and other constraints. The majority of these limitations affect Earth-observation CubeSat-class spacecraft due to the lack of high-performing attitude determination and control systems. Achieving accurate and robust attitude determination and control system in small form factor is challenging and complete attitude knowledge in inertial space requires a combination of multisensors in order to maintain robust attitude estimation in different cases. Our objective in this dissertation is to improve attitude estimation of Earth-observation satellites using visual and inertial sensors with Unscented Kalman filter. CMOS （Complementary Metal-Oxide Semiconductor） camera is common sensor for CubeSats to carry as its primary payload. Besides, Earth’s surface features in an imager’s field of view provides an important prior feature for attitude estimation. These observations motivate us to use a CMOS camera as sensor for small satellite’s attitude determination system. Our goal is to improve the accuracy of attitude estimation capabilities of the resource-constrained small satellite by using a sensor fusion approach with vision-based and inertial sensors. This improvement increases the possibility of applications that require accurate attitude determination and control system such as laser communications, target tracking and Earth imaging missions for CubeSats. In this dissertation, we study the issues of vision-based and inertial-aided attitude determination approaches and sensor fusion techniques. The vision-based attitude estimation method is usually achieved by Earth’s surface features tracking and featurecorrespondences matching techniques with different outlier rejection methods and then provides absolute attitude information in 3 degrees of freedom. Experimental and simulation tests were carried out to qualify the proposed approach’s performance. For the validation of the vision-based method’s performance, we used raw imagery data of High Definition Earth Viewing （HDEV） payload of the International Space Station （ISS）. The performance of the visual-inertial approach is assessed through the realistic Earth-surface scene simulations and results are compared with ground truth data. All the methodology details of the proposed approach have been explained and the test results have been analyzed., 九州工業大学博士学位論文 学位記番号：工博甲第483号 学位授与年月日：令和元年9月20日, 令和元年度},
 school = {九州工業大学},
 title = {Visual-Inertial Attitude Determination for Small Satellites},
 year = {}
}