Spacecraft Attitude Estimation Using Adaptive Gaussian Sum Filter Jemin George * Gabriel Terejanu † and Puneet Singla ‡ ABSTRACT This paper is concerned with improving the attitude estimation accuracy by implementing an adaptive Gaussian sum filter where the a posteriori density function is approximated by a sum of Gaussian density functions. Compared to the traditional Gaussian sum filter, this adaptive approach utilizes the Fokker-Planck-Kolmogorov residual minimization to update the weights associated with different components of the Gaussian mixture model. Updating the weights provides an accurate approximation of the a posteriori density function and thus superior estimates. Simulation results show that updating the weights during the propagation stage not only provides better estimates between the observations but also provides superior estimator performance where the measurements are ambiguous. INTRODUCTION The spacecraft attitude estimation problem involves determining the orientation of a spacecraft from on-board ob- servations of line-of-sight vectors to various reference points such as celestial bodies, the direction of the Earth’s magnetic field gradient, etc. [1]. Generally, a redundant set of these observations is used to generate more accurate estimates of the spacecraft attitude. Several attitude sensors are discussed in the literature, including three axis magne- tometers, Sun sensors, Earth-horizon sensors, global positioning sensors, rate integrating sensors and star-cameras [2]. Accuracy of the estimated attitude depends on the quality of attitude sensors used. The estimation algorithm is re- quired to extract the useful information from the available sensor measurements, which are often corrupted by sensor noise, biases and sensor inaccuracies. Generally, the attitude estimation algorithms can be divided into two categories: 1) batch attitude estimation algorithms such as TRIAD, 3 QUEST, 4 ESOQ, 5, 6 etc and 2) sequential attitude estimation algorithms such as Kalman filter, 7 REQUEST, 8 un-scented Kalman filter, 9 particle filter, 10 etc. A detailed discussion on various attitude estimation algorithms can be found in Ref. [11]. During normal operations, the attitude estimation problem needs to be solved recursively; i.e., the attitude filter makes new updates and predictions based on present and prior sensor information. Though many recursive attitude estimation algorithms are presented in the literature, * Graduate Student, Department of Mechanical & Aerospace Engineering, University at Buffalo, Buffalo, NY-14260, Email: jge- orge3@buffalo.edu. † Graduate Student, Department of Computer Science & Engineering, University at Buffalo, Buffalo, NY-14260, Email: tere- janu@buffalo.edu. ‡ Assistant Professor, AIAA, AAS Member, Department of Mechanical & Aerospace Engineering, University at Buffalo, Buffalo, NY-14260, Email: psingla@buffalo.edu. 1