Robust Control for Underwater Cooperative Localization Systems With Unknown Noise and Multiple Nodes

Abstract

The cooperative localization technology of multi-AUV has been a research hotspot in the area of underwater localization in recent years. Aiming at the situation that the localization accuracy is reduced and diverging due to uncertain information such as outliers and unknown time-varying noise in the measurement information in the dynamic topology environment, a new measurement filtering module is designed by using factor graphs. First, the K-means algorithm is introduced based on the cooperative localization algorithm based on factor graphs in the dynamic environment, and an outlier filtering module is designed to filter measurement outliers. After outlier correction, the EM algorithm is introduced to resolve the issue of unknown measurement noise variance caused by unknown time-varying noise and design an adaptive filtering module. Finally, to address the issue of degraded real-time performance due to a large number of system communication nodes and diverse quality, the paper introduces the CRLB algorithm, distance evaluation factor, and measurement smoothing factor to design a node optimization module. Finally, these modules are integrated into the cooperative localization system through factor graphs. Through experimental analysis, the proposed algorithm effectively reduces outliers and unknown time-varying noise in measurement information, as well as the impact of a large number of system nodes on localization accuracy and real-time performance. In addition, the proposed algorithm also shows strong robustness in the face of extreme underwater environments.