Adaptive Unscented Particle Filter Algorithm for Acoustic Positioning of AUV Docking in Shallow Water

Abstract

The positioning of Autonomous Underwater Vehicle (AUV) plays an important role in AUV docking. However, in the shallow water, performance of acoustic devices is significantly limited to ambient noise that leads the generation of positioning outliers which seriously affect the AUV docking guidance and control. In this case, an appropriate filtering method leverages the smooth and consecutive state observation that efficiently eliminates the outliers and leaping data. In this paper, an Adaptive Unscented Particle Filter (AUPF) based on the Unscented Particle Filter (UPF) algorithm framework and the thought of Strong Trace Filter (STF) is proposed. By testing several filters on the outline acoustic positioning data set, it is noticed that the performance of AUPF is better than Particle Filter (PF) and UPF compared with the high-precision location data of Inertial Navigation System (INS) which is used as the reference. The experiment shows that the suggested filter can successfully remove outliers generated by ultra-short baseline (USBL) acoustic positioning systems in shallow water, correct AUV positioning data, and increase navigation accuracy.