The nine-axis MEMS Inertial Measurement Units (IMU) have been widely used in various fields, such as underwater vehicles, unmanned aerial vehicles and bionic robots. Due to the noises of gyroscope sensors and errors introduced in solution process, the rotation angles estimated using only angular velocity data usually contain large accumulated errors and have to be corrected by acceleration and geomagnetic measurements. A serious problem is if there is strong magnetic anomaly field in the environment, the geomagnetic field aiding performance decrease quickly and probably leads to extra errors. To improve the heading and attitude estimation accuracy of the nine-axis MEMS IMU in magnetic anomaly field, a partially adaptive Extended Kalman Filter (PADEKF) using double quaternions is proposed in this work. To reduce the coupled influence of magnetic measurement noise on attitude estimation in a single quaternion, the heading and attitude angles are represented with two independent quaternions in the state vector. Self-adaptability design is adopted in the EKF to improve the robustness to spatially varying magnetic anomaly data. For the case that the strong and quickly varying magnetic anomaly field cannot be well modelled by the PADEKF, a combination algorithm of long and short-term memory (LSTM) neural network and Runge-Kutta method is given to make good heading estimation. Field experiments in different scenarios are performed and verified the effectiveness of the proposed approach.

Due to the position drift of inertial navigation systems, it is still challenging to achieve long-term and accurate position estimates during underwater navigation. The seabed topography has been proven to be effective in aiding information for accurate positioning benefiting from its rich spatial variation. With the advantage of the multibeam echosounder (MBES) in efficient bathymetric survey, the simultaneous localization and mapping (SLAM) approach can be performed using bathymetric data in unknown environments for underwater vehicles to get good position estimates. The SLAM performance relies on the number and accuracy of loop closures heavily. Thereby, the capabilities of the data association method and solver in dealing with the uncertainties of vehicle pose estimates, bathymetric data, and topographic features affect the SLAM performance strongly. This work proposes a new graph-based bathymetric SLAM method to improve the robustness of the uncertainties in both factor-graph construction and optimization stages. In the front end, on the base of a matching suitability-based MBES submap construction method, a dual-stage bathymetric point cloud registration approach that is able to detect most false loop closures is proposed. In the back end, a robust optimizer based on Frechet distance is introduced to further identify and remove the false loop closures missed in front end. Experiments using field MBES bathymetric data sets are conducted to verify the effectiveness of the proposed approach.

This article proposes a new method to simultaneously estimate the locations and magnetic moments of multiple magnetic dipole sources without the prior knowledge of the number of dipoles in the 3-D detection region. By initializing a large number of dipole sources evenly spaced in the detection region as potential candidates for the true dipoles, we introduce an indicator parameter for each dipole candidate such that its Sigmoid function is the probability that the candidate converges to a true dipole. A joint optimization is then formulated to minimize the mean square of the regularized error between the measured magnetic gradients and the calculated gradients from the estimated dipoles. The proposed nonlinear optimization is solved by the Levenberg–Marquardt algorithm, yielding the indicators and their corresponding dipole locations and magnetic moments. The implementation details are also provided, such as using multiple initialization schemes to avoid local minima, selection of measurement points and candidate locations to avoid the “high-wall effect,” and the need for preprocessing measurement data to avoid interference. Extensive simulations are conducted to investigate the effects of parameters, noise, and interference on the detection performance, and the results show that the proposed algorithm is robust in different scenarios as long as the total number of measurements is larger than the total number of unknowns in the optimization problem. When the false alarm rate is set at $5\times 10^{-2}$ , the proposed algorithm achieves $Recall$ of 0.91, 0.86, and 0.78 for the number of true dipoles being $N=2,4$ , and 6, respectively. The performance is robust against external interference and parameter selections.