Application of a full-dimensional observable smoothing algorithm in SINS

The traditional zero-velocity correction algorithm (ZUPT) can theoretically suppress the accumulation of navigation errors, but it can only correct the state errors of the zero-velocity interval and the observations are less. The state information of the model solution based on the SINS algorithm and the navigation error equation in the non-zero-velocity interval causes a sudden change in the entire pedestrian trajectory process, that is, the stability is poor and the accuracy is not high. This paper proposed a algorithm, which 3D errors of attitude and position are added as observations to achieve full-dimensional observability based on the traditional zero-velocity correction, then all state errors are obtained by EKF estimation, and a post-processing smoothing algorithm is introduced to make full use of the measurement information over the entire time period to correct the navigation errors of the non-zero-velocity interval. In order to verify the accuracy and stability of the algorithm, experiments were carried out by using the self-developed IMU. The results show that the proposed algorithm has better stability than the traditional ZUPT, improves the smoothness of the trajectory, and the navigation accuracy is improved by 1.7%.