A Long-Term Underwater Robust Terrain Contour-Aided Navigation Method With a Single Beam Echo Sounder
Dong Ma, Teng Ma, Ye Li, Yu Ling, Yueyang Ben
Abstract
Terrain-aided navigation (TAN) system implemented with a single beam echo sounder (SBES) has shown potential for long-term accurate navigation aboard autonomous underwater vehicles (AUVs), without external navigational aids in the form of acoustic or satellite positioning methods. However, the SBES with limited observational capability makes the accuracy of a TAN system with a SBES highly dependent on the resolution of an a priori map, and the number of terrain features. To achieve stable navigation information even in areas with the above-mentioned limitations, this article proposed a robust terrain contour-aided navigation method to yield an accurate vehicle state estimate for long-range AUVs. A particle filter-based approach is proposed to estimate the vehicle's position by fusing navigational results from both TAN and the inertial navigation system (INS). A contour map is applied as an a priori map instead of a traditional digital elevation model (DEM) to improve the performance of navigation under a low-resolution map and poor terrain features. To deal with the possible invalid TAN results under smooth terrain, an algorithm based on the terrain contour confidence lower bound is proposed, which alters the re-sampling process in smooth/complex terrain. Play-back experiments with a 25-hours-length field data collected from an at-sea experiment have shown that the proposed method can realize robust and accurate long-term navigation for an AUV in both smooth and complex terrain. Compared to the traditional TAN method, the average positioning error and maximum positioning error have been decreased by 62.9% and 48.1%, respectively. The proposed method can also maintain robust navigational results in play-back experiments even with a 50 m resolution a priori map or 150 m initial positioning error.