A multi-AUV cooperative navigation method based on the augmented adaptive embedded cubature Kalman filter algorithm

Cooperative navigation is part of the essential method for multiple autonomous underwater vehicles (AUVs) to gain an accurate position when performing tasks underwater. However, the positioning accuracy and stability of the traditional algorithms are affected by the dimension of state-space and the unknown time-varying noise in the ocean, which cannot meet the demand of the increasing positioning performance. For this problem, we propose a cooperative navigation method based on the augmented adaptive embedded cubature Kalman filter (A-AECKF) algorithm. Due to the non-additivity of the system noise in the realistic model of multi-AUV cooperative navigation, we augment the dimension of the state variables firstly, which combines with the system model to estimate the noise. Then we adopt embedded cubature criterion reselecting cubature points and their weights to minimize the positioning error caused by state augmentation. Finally, a nonlinear noise statistical estimator is used to estimate the time-varying ranging noise in real time, which effectively suppresses the positioning error caused by the non-Gaussian white noise. We evaluate the performance of the proposed A-AECKF cooperated navigation method comprehensively under the circumstance with high-dimensional state-space and the unknown time-varying measurement noise. Compared with other related algorithms, the experimental results indicate that the presented method possesses higher positioning accuracy and stability.