The underwater terrain-aided navigation (TAN) method with a single beam echo sounder (SBES) can achieve long-term underwater navigation. However, due to the limited bathymetric measurements collected using a SBES, the accuracy and robustness of the TAN method are affected by a variety of factors, such as the richness of terrain features and the gradient of underwater terrain. Consequently, the SBES-TAN methods with gridded or contour priori maps cannot yield robust navigational results for long-range autonomous underwater vehicles (AUVs). To address this issue, this paper proposed a robust fusion TAN (RFTAN) method with a SBES on the basis of particle filter (PF) theory to locate a vehicle accurately. In the proposed method, both terrain positioning results provided by gridded-TAN and contour-TAN were fused to weigh a particle, with the fuzzy-theory-based fusion weight division (FWD) method. Playback experiments were conducted using 25-h-length field data collected from an at-sea experiment, and experimental results show that the proposed method can provide both robust and accurate long-term navigational results for AUVs.
Read full abstract