Improved AUV navigation algorithm fusing SINS/DVL/PS and considering lever arm errors

Abstract Autonomous underwater vehicles (AUVs) play an increasingly important role in underwater operations, and continuous and stable high-precision positioning, navigation, and timing (PNT) information serves as the foundation for their optimal performance. On an AUV, the strap-down inertial navigation system (SINS), Doppler velocity log (DVL), and pressure sensor (PS) are common PNT sensors. To obtain highly precise and reliable navigation results, it is necessary to fuse the measurements of these three sensors. However, there is spatial inconsistency in the reference centers of them. In addition, there is a difference in height anomaly between the starting surface of depth (quasi-geoid) and the starting surface of geodetic height (reference ellipsoid) in the depth constraints. If these factors are not considered during fusion processing, the accuracy of the navigation solutions will decrease. To address the above issues, we present an adaptive filtering method with lever arm errors and height anomaly compensation for SINS/DVL/PS integrated navigation. Firstly, we introduce PS to constrain the height information of the underwater vehicle. Following that, during the depth constraint, when converting the depth observed by PS into geodetic height, the height anomaly correction is taken into account. At last, the measurement equation of SINS/DVL/PS integrated navigation considering lever arms among the three sensors and height anomaly are derived in detail. To verify the algorithm, AUV field experiments near the coast were conducted. The results show that the proposed method outperforms the method without considering lever arm errors with an improvement of about 10% in the planar direction of navigation accuracy. To further verify whether a larger lever arm has greater impact on accuracy, simulation experiments were designed. The results reveal that the errors increase with the lever arm values and there is a more significant improvement in accuracy after accounting for lever arm errors. These findings can provide theoretical basis for SINS/DVL/PS integrated navigation.