Attitude Estimation Employing Common Frame Error Representations

This work studies vector representation in the attitude filtering problem. Because vector coordinates must be estimated in lieu of the abstract vector itself, the choice of representation is important. A challenge arises in attitude estimation when the coordinates of a vector are sought with respect to the attitude state coordinate frame. Although a wealth of research examines filtering with attitude state constraints or discontinuities, this work focuses on vector state estimation with respect to the attitude state frame. Specifically, an alternative vector state error is defined using common coordinates over all vector error realizations. A modified extended Kalman filter is developed with this error realization as its foundation. Numerical simulations are conducted, and results obtained using the new filter are compared with those generated by employing a standard estimate error definition.