On improved Simultaneous Localization and Mapping algorithm for underwater navigation

摘要

Improved Simultaneous Localization and Mapping (SLAM) algorithm for underwater navigation here is based on the underwater robot Autonomous Underwater Vehicle (AUV) as a carrier. This paper focuses on the Extended Kalman Filtering (EKF) algorithm, which can solve problems of SLAM. However, the EKF algorithm generates modeling errors in the process of linearization, and also has unknown modeling errors in the state transition matrix and observation matrix. In order to solve those problems, the paper presented the EKF algorithm based on virtual noise compensation technology. Simulation platform built on the AUV system model can test the improved algorithm via filtering accuracy, convergence and stability. Simulation results showed that, compared with the traditional EKF algorithm, the improved EKF algorithm with virtual noise compensation technology can significantly improve the performance of nonlinear filtering, and solve the accuracy issues of SLAM.