INS/DVL Integrated Navigation By Using Backtracking Scheme

Inertial Navigation System(INS) combined with a Doppler Velocity Log(DVL) is a common approach for Autonomous Underwater Vehicles(AUV) navigation. However, as the velocity provided by DVL is in the Doppler instrumental frame and the lack of external information, there are still following challenges in calibration of DVL, fast INS initial alignment and integrated navigation:1 The accuracy of the calibration is affected by INS initial alignment, and vice versa.2 In the case of only the velocity of DVL is available, fast initial alignment is difficult due to the low observability of the heading.3 The alignment time is limited. In the case of the in-motion alignment, there will be large misalignments if coarse alignment is short. However, if the coarse alignment occupies too much time, it is difficult for the fine alignment to achieve the required accuracy.4 Once the positioning observations are not available, parameters estimations and compensation of the position error remains a difficult problem.In order to solve the problems mentioned above, the main contributions and innovations of this paper are presented as follows:1 A backtracking method for the attitude is proposed to eliminate the effect of the INS initial alignment in calibration of DVL’s velocity. It is able to improve the robustness of the calibration in the case of the violent environments. With the proposed method, the AUV only needs to arise for one time to finish the online calibration.2 An optimization-based INS initial alignment model and algorithm which uses DVL aiding is proposed. There are no positioning observations in the case of DVL aiding. With a backtracking scheme, the position of the AUV can be obtained during the alignment and thus the alignment model is established. Meanwhile, compared with other sampling strategies, the interleaved sampling strategy is employed to improve the accuracy of the alignment.3 A Kalman-based backtracking INS alignment model and algorithm using DVL aiding is proposed. The heading error shows slow convergence in DVL-aided alignment for its low observability. In order to meet the requirement of the fast INS initial alignment, a novel backtracking scheme is employed. During the process of the coarse alignment, the data that will be used for the subsequent fine alignment is recorded. Therefore, the fine alignment occupies no extra time and thus will speed up the overall alignment process. In addition, it lengthens both the coarse and the fine alignment time and hence improve the accuracy of the alignment. Due to the high update rate of the INS, a large volume of the data has to be recorded by using the traditional fine alignment model. In this paper, a novel INS initial alignment model which is derived in the inertial reference frame is proposed to reduce the volume of the data that need to be recorded during the coarse alignment and hence make the proposed method possible for real applications.4 By using Unscented Kalman filter(UKF), a large misalignment model for DVL-aided INS initial alignment is proposed. This method is compared with the two alignment approaches mentioned above on convergence time, accuracy and robustness.5 An integrated navigation and parameters estimation model using backtracking scheme is proposed. The backtracking alignment model mentioned above is extended to the navigation. The sub sequential position can be obtained in the case of only the initial position and the velocity of DVL is available. The initial alignment and navigation can be done in the backtracking process by recording only a few data. This technique can be used in fast lauch. In addition, the INS biases are estimated to improve the performance of the navigation. It is shown by the experimental results that the INS biases can be estimated reliably by this new error model and hence improved the performance of the navigation. In addition, the accuracy of the navigation by using this method is not lower than that with the traditional model. And the AUV can be lauched without waiting for the initial alignment.