ATTITUDE DETERMINATION USING GPS VECTOR OBSERVATIONS

The vectorized algorithms for GPS based attitude determination can be equivalent to a two-level optimal estimation problem. In the first level estimation the GPS carrier phase measurements can be converted to the vector observations in a sense of least squares. The second level estimation is the Wahba’s problem which describes how the attitude matrix can be derived from the vector observations. In contrast to the traditional algorithms resolving the Wahba’s problem through the quaternion approach, e.g. the QUEST method, this paper presents an iteration method to resolve the Wahba’s problem through the small-angle approach. The two-level optimal estimation turns out to be globally optimal under the so-called balanced constellation condition or the balanced baseline condition. The experiment results show that the iteration method can produce the solution of the same accuracy as the solution derived from the QUEST algorithm. The improved TRIAD algorithm is also presented and evaluated in the experiments.