|Institution:||Delft University of Technology|
|Keywords:||Microsatellite; Attitude; Estimation; Determination; ESEO; Maximum Information Rate|
|Full text PDF:||http://resolver.tudelft.nl/uuid:3ba10d0d-c76b-4f32-80df-d0704bd7c963|
The European Student Earth Orbiter (ESEO) is the main educational project of the European Space Agency that allows students to design, develop and test scientific payloads. The contribution to be made by the TU Delft is a software-based Attitude Determination Experiment (ADE) that contains four different algorithms for attitude estimation that are to be tested in-situ for comparative analysis through the telemetry sent down by the satellite. The four algorithms are the Optimal REQUEST, the Additive Quaternion Kalman Filter, the Pseudolinear Quaternion Kalman Filter and the Multiplicative Extended Kalman Filter. The Additive Quaternion Kalman Filter has been further adjusted with the incorporation of the Maximum Information Rate Filter, which reduces the measurement matrix into six candidates and selects the one corresponding to the highest information rate for use in the algorithm. Though this filter has the benefit of using reduced matrices, computational efficiency is only really increased if the selection process is as computationally lean as possible as well. Therefore, an analysis was performed of the selection pattern of the Maximum Information Rate filter. This analysis shows that a clear pattern exists for the Earth sensor (based solely on the hemisphere), a more complicated pattern is visible for the Sun sensor, while the magnetometer has no useful pattern to speak of. The ADE has been programmed in the C-language and adjusted for the RTEMS real-time operating system. All operations to be performed by the payload have been verified to operate within the assigned budgets and limitations of the hardware on board the satellite. The payload is capable of running the four algorithms within 100 milliseconds while retaining a code size 24 kB, which is lower than the 35 kB budget and a memory size of 8.8 kB, which is lower than the 10 kB budget. All algorithms have undergone extensive Monte-Carlo simulations to test for stability and sensitivity to the involved parameters. All algorithms manage to retain a steady-state angular estimation error lower than 0.5 degrees under conditions varying by about 30% from the expected parameters. At the time of writing, the ADE has passed the Critical Design Review and is ready to be integrated onto the ESEO satellite. Advisors/Committee Members: Kuiper, J.M..