Nonlinear state-estimation for spacecraft attitude determination

Spacecraft requirements are changing. Smaller, lower-cost spacecrafts are more desirable, while the desired performance is increasing. The effect of using cheaper hardware may be offset by the use of more sophisticated controller and estimation schemes.; This thesis compares the Extended Kalman Filter (EKF) with the Nonlinear Predictive Filter (NPF) and the extended H_∞ filter (EHF). The EHF is a linearized H_∞ filter, completely analogous to the EKF, which is a linearized Kalman filter.; Under certain circumstances, the NPF outperforms the EKF. However, the requirements for these circumstances conflict with filter robustness with respect to initial estimate error. This problem is solved by applying the NPF differently, but comes at the cost of robustness with respect to plant error. The EHF performs poorly in all aspects.; The EKF has the least sensitivity to sensor noise level and is the estimation scheme of choice.