A new strapdown attitude algorithm

This paper develops the application to the strapdown attitude problem of the rotation vector concept, in which obtaining a solution for the rotation vector and updating the attitude quaternion are considered entirely separately. A new rotation vector algorithm is derived which takes three samples of gyro data per update, and offers greatly improved coning performance. This is obtained by making the assumption that gyro output varies according to a third-order time relationship, rather than second or first order as in other algorithms. Performance of the three sample algorithm is compared with well-known algorithms in a coning environment and in a random motion environment. The quaternion updating may be performed to whatever degree of accuracy is required: An economical modified fourth-order algorithm is proposed, and its comparative performance shown. This approach to the problem gives versatility in that computing resources use may to some extent be tailored according to the relative extents of coning and of fixed axis rotation in the expected environment.