Attitude estimation of a small spinning satellite using Kalman filter approaches

Nanosatellites provide a low-cost solution to designing, developing and building space systems. However, there are several constraints linked to the low-cost, compact size and quick development time of such nanosatellites. These cost constraints restrict the use of space-grade reliable hardware. These factors lead to the risks pertaining to the reliability and robustness of the system. The Attitude Determination and Control System (ADCS) is one of the subsystems for which both the hardware and the software are affected by the adverse effects of the closely integrated subsystems. The main objective of this thesis is to study more practical solutions for the attitude estimation algorithms of a fast spinning cubesat using Kalman filter approaches. The aim also encompasses the utilisation of a single algorithm for both the 3-axis and spin-stabilised attitude modes while also considering the computational load. The design of the Aalto-1 and Aalto-2 onboard computers (OBC) has been presented and discussed. The design focus for Aalto-1 and Aalto-2 OBCs has been system safety and robustness in the space environment. A cold redundant system topology has been used for the Aalto-2 OBC which also hosts the ADCS algorithms. The ADCS algorithms lie at the core of the whole ADCS system. Small satellites ADCS face several challenges such as the limited computational resources, poor sensor noise characteristics due to cost constraints and sensor biases due to EMI etc. These factors have to be considered during the algorithm design and implementation. Due to this, the biases also need to be estimated and the sensors must be calibrated during the flight. The algorithms have been designed and tested using the Plasma Brake Experiment (PBE) scenario. The PBE is the deorbiting experiment for which The satellite needs to be spun up to 200 deg/s. It will be hosted onboard Aalto-1. The spin controller requires the attitude estimate for its operation. The controller performance has been observed to improve with greater accuracy in the attitude estimates. One way of improving the estimation accuracy is to use online parameter estimation. The adaptive Unscented Kalman Filter (AUKF) has been tested for the PBE spin phase. The results are encouragingly in favour of the AUKF as compared to the UKF. However, the UKF has been observed to be complex and computationally expensive for a nanosatellite. Therefore, an alternate solution based on the Pseudo-Linear Kalman Filter (PSLKF) has been studied for PBE. The PSLKF has been designed and studied with an attitude-independent magnetometer calibration algorithm. The PSLKF solution has been found to be computationally lighter with an effective magnetometer bias estimation algorithm. Although, the bias estimation convergence is slower, the overall estimation results are competitive as compared to the UKF