A Rodrigues-vector-based strapdown attitude algorithm

Abstract

A new class of Rodrigues-vectors-based strapdown attitude algorithms is developed in this paper. This class of algorithms uses Rodrigues vectors instead of the widely used rotation vector and applies the conception of least square technique. The multiple summations of gyro samples is used to obtain the least square estimation of Rodrigues vectors increment in an attitude update period. Such estimation is obtained by assuming that gyro output can be approximated by algebraic polynomial of time in an update period. The global attitude Rodrigues vector (or attitude quaternion) is updated by multiplying the previous attitude with the computed attitude Rodrigues vectors increment. Theory analysis and numerical simulation show that the proposed algorithms can effectively suppress the non-commutativity error and have high computational efficiency. In addition, the proposed algorithms have simple sequential executive structure. This class of strapdown attitude algorithms provides a good balance among accuracy, computational burden and structure complexity. It is easily adapted to various applications.