A reliable and fast lost-in-space mode star tracker

• Main Author: Metha Deval Samirbhai
• Other Authors: Chen Shoushun
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Aeronautical engineering::Aircraft; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision; DRNTU::Engineering::Aeronautical engineering::Electrical systems and equipment; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Pattern recognition
• Online Access: https://hdl.handle.net/10356/88529; http://hdl.handle.net/10220/47620
• Institution: Nanyang Technological University

Developing an autonomous and reliable attitude determination system is one of the most crucial requirements for a satellite mission in space. Depending upon the satellite missions, the demand of the attitude (orientation) accuracy can be as high as 1 arc second. Thus, it is necessary to develop a highly reliable and autonomous attitude determination sensor for such satellite missions. The central theme of this thesis is to develop such an attitude determination sensor - a star tracker. A star tracker operates in two modes - Lost-in-space (LIS) mode and tracking mode. In LIS mode, it has to establish the initial attitude without any prior information. When the satellite is initially placed into the orbit or when it loses the attitude information due to some problem, the star tracker operates in the LIS mode to establish the attitude information. In the tracking mode, the star tracker obtains the attitude information from the LIS mode and then merely updates the attitude based on the subsequent images of the stars captured. Thus, LIS mode is critical to the reliable operation of the star tracker. In the LIS mode of the star tracker, the process of reliable attitude calculation depends upon identifying the stars correctly in the image captured, also popularly known as star identification or star pattern recognition in the research community. A star identification algorithm is expected to provide a high identification accuracy, robustness, and low time complexity in the scenario of patch mismatch, magnitude uncertainty, positional deviation, and false stars present in the image captured. To achieve both high robustness and low time complexity has been a challenge for the researchers in the past two decades. In the research work pursued in this thesis, firstly we analyze the problems faced - patch mismatch, magnitude uncertainty, positional deviation, and false stars in achieving a reliable star identification. We provide a quantitative analysis with a deep understanding of the gravity of the problems mentioned above. Having analyzed these problems, we develop a framework for achieving high robustness and low time complexity for the process of star identification. We propose three novel approaches for solving the problem of star identification. For initial testing of the developed approaches, we produce simulated star images (containing the problems faced) which closely resemble the real star images. Later, we implement the proposed methods on hardware and develop a LIS mode star tracker prototype. We configure a state-of-the-art star tracker hardware-in-loop testing system for testing any star tracker prototype. Finally, we test the performance of the LIS mode star tracker prototype on the real images captured by a star tracker SST-20S currently mounted on a satellite VELOX-CI. The high accuracy of orientation provided by the star trackers has made them successful to be adopted as attitude determination sensors for satellite missions. Making them reliable is a challenge, which we have addressed in the research work pursued in this thesis. Star trackers can also be adopted as attitude sensors for the emerging nano and pico satellites. In this thesis, we have identified a few research gaps faced in the process of star pattern recognition. We believe that the framework, techniques, and algorithms developed for LIS mode star tracker in this thesis will assist the research community to achieve better performance for attitude determination of satellites.