Research on Cooperative Localization of Multi-UUVs Under Incomplete Observation

Abstract

Localization is the basis of unmanned underwater vehicle (UUV) to carry out missions, and cooperative localization of multi-UUVs has been a research hotspot in recent years. Packet loss, outliers, and unknown noise are among the challenges cooperative localization encounters underwater. These challenges can be summed up as "incomplete observation" problems. To tackle the problems, firstly, the extended Kalman filter (EKF) algorithm is used to achieve data fusion while accounting for packet loss. Secondly, M-estimator is applied to EKF to improve its robustness against outliers, resulting in a robust extended Kalman filter (REKF). Thirdly, the Sage-Husa algorithm is used to improve the adaptiveness against unknown noise of EKF, allowing it to estimate the statistical properties of observation noise and resulting in an adaptive extended Kalman filter (AEKF). Naturally, the three methods are combined as a robust adaptive extended Kalman filter (RAEKF) to simultaneously resolve the three incomplete observation problems. Finally, MATLAB is used to run simulation experiments to test the algorithms’ performance, the superiority of REKF and AEKF is highlighted by comparing with EKF under single problem, and the superiority of RAEKF is highlighted by comparing with REKF and AEKF under combined problems. This study has positive implications for the application of cooperative localization of Multi-UUVs.