Improved decentralized attitude estimates using the covariance intersection algorithm
MetadataShow full item record
This thesis looks at achieving better attitude estimates by using the Covariance Intersection algorithm to fuse state estimates and covariances from various sources. The attitude parameterization of choice is the quaternion. Because of the norm constraint on the quaternion, the Covariance Intersection algorithm is derived from an augmented optimization problem using the method of Lagrange multipliers to handle the constraint. When state vectors include only a single quaternion, several analytic solutions to the resulting CI equations exist. For state vectors containing two or more quaternions, a multi-variate Newton-Raphson algorithm is developed. Two simulations are presented to show the advantage of the Covariance Intersection algorithm as applied to spacecraft attitude estimation. In each case, results from the Covariance Intersection algorithm are superior to those attained using simple extended Kalman filters.