Fast observers for spacecraft pointing control

This paper deals with designing fast stochastic observers for spacecraft pointing control. The motivation is to try to circumvent the long settling times associated with the optimal Kalman filter, which are often on the order of several hundred seconds in many typical spacecraft applications. For this purpose, the Karush-Kuhn-Tucker necessary conditions are used to solve for an optimal constrained stochastic observer which minimizes the variance of the attitude estimation error subject to the constraint that its poles lie to the left of a specified vertical line in the Laplace s-plane. A globally optimal analytical solution is found for the typical case where the attitude estimator is comprised of three decoupled single-axis observers, each of second-order form