Efficient Terrain-Aided Navigation System with Submap-Size Calculation and Embedded System

Abstract Underwater terrain-aided navigation (TAN) holds high potential for long-term underwater accurate navigation of autonomous underwater vehicles. TAN can locate a vehicle by calculating the similarity between an a priori map and a vehicle's real-time observation consisting of a set of bathymetric measurement points. However, the amount of measurement points in the real-time observation affects both positioning accuracy and computational consumption of the TAN system, making it challenging to calculate a suitable size of real-time observation in TAN. With a smooth seabed terrain, a small observation area leads to insufficient topographic features and finally an inaccurate matching result, while a large area with a mount of features results in high computational cost. This paper proposes a method to restrain the size of observation in TAN systems based on terrain entropy and difference of normals. Meanwhile, this paper implements the TAN algorithm into an embedded system architecture used by actual underwater vehicles that are already in service to reduce the power consumption of the TAN system. The effectiveness of the algorithm has been demonstrated through playback experiments based on a semi-physical simulation platform using a PC/104-embedded computer.