Research on Performance of Terrain Matching Algorithm for Underwater Autonomous Vehicle Based on Particle Filter

Abstract

AUV is a powerful tool for exploration and development of marine resources. Underwater terrain matching navigation is an emerging passive autonomous navigation method, which improves the long-term and high-precision navigation capability of AUV. The underwater terrain matching algorithm based on particle filter (PF) use a large number of particle sampling to approximate the system state, which can avoid the Taylor expansion truncation error caused by the terrain linearization method, and has good matching precision and terrain adaptability. However, due to the strong nonlinearity and the uneven distribution of the underwater terrain features, it is difficult to achieve the full-range terrain matching optimize with a single configuration parameter of PF underwater terrain matching algorithm. For this reason, this paper has comprehensively studied the PF terrain matching algorithm (referred to as PF matching algorithm). The principle, accuracy and implementation steps of PF algorithm are analyzed. With the UKF algorithm as a reference, the advantages and disadvantages of PF-based algorithm are studied through computer simulation of terrain matching navigation. The matching precision and terrain adaptability of PF-based algorithm has been tested. From the simulation results, it is concluded that the particle diversity decline is one of the main factors affecting the matching performance, which provides a basis for algorithm performance improvement and engineering adaptability improvement.