A Parallel Attitude-Heading Kalman Filter Without State-Augmentation of Model-Based Disturbance Components

This paper proposes an inertial and magnetic sensing-based parallel orientation Kalman filter (KF), which is composed of two separate KFs: one for attitude estimation with an external acceleration model as presented in author's previous work and the other for heading estimation with a magnetic disturbance model. The novelty of the proposed attitude-heading KF, compared to other model-based KFs, is that the disturbance models are implemented by inserting them into the measurement equations without state-augmentation. In this case, there is no increase in the matrix size of the corresponding KF structure, which is advantageous as the computational cost is approximately halved when compared with the state-augmentation type approaches. The proposed KF has been experimentally verified in magnetically disturbed conditions. The proposed method has advantages over the KF with state-augmentation in the calculation efficiency and ease of use.