Novel Attitude Estimation Of Strapdown Inertial Navigation Systems With Singular Value Decomposition Technique

Abstract

Davenport’s q method & the Singular Value Decomposition (SVD) method are the two vigorous estimators that reduces Wahba’s loss function. In these, the q method is slightly quicker due to its computation of optimum quaternion as an eigenvector of a symmetric 4x4 matrix through the prevalent eigenvalue. The ESOQ and ESOQ2 (EStimators of the Optimal Quaternion) and the QUEST (QUaternion ESTimator) algorithms are less determined as the extreme eigenvalue’s distinguishing polynomial equation is solved by them. These estimators are apt to track the undulations of the sea with equivalent precision and accurateness. The SVD method is chosen and shown to be the most robust of all the hostile methods for the orientation of SDINS (Strap-Down Inertial Navigation Systems) using rate matching observations at sea in this paper. SVD is known most robust decomposition of all the decompositions of a matrix. SVD based attitude estimation being a batch technique would suffer from much less computational issues.