Research On The Simultaneous Localization And Mapping Technology Of AUV

As the growth of the population, land resources available has become less and less, the line of sight of people are turning to the ocean, but sea world pressure big, low visibility, conditions and other issues, people cannot easily into the ocean world. To solve this problem, people began to develop a new type of Autonomous Underwater robot (Autonomous Underwater Vehicle, the AUV). But, the AUV has defects:If there are not specific environmental information, an underwater robot is unable to conduct a series of autonomous tasks:underwater operations, positioning, path planning, decision-making. And the environment is extremely complex.with much interference, noise, if just by virtue of the internal sensors, it is unable to achieve high-precision navigation technology in the underwater environment. Simultaneous localization and map building technology not only include external sensors and use these to perceive environmental information, but also it can be merge the information form the internal and external sensors to build an underwater robot to locate and map. Simultaneous localization and mapping underwater environment does not require the priori information, it can be extracted and updated underwater robot, positioning information and external environment information through;the establishment of the movement and observation models. With the continuous development and attention of the AUV in the marine sector,SLAM technology in underwater navigation has also become a hot spot.Currently, In order to achieve the AUV simultaneous localization and map building technology environment, the Extended Kalman filter algorithm become the broadest algorithm. However, with the in-depth study of the extended Kalman filter algorithm, the researchers found many of the limitations of the Kalman filter algorithm:first, the basic principle of EKF is linear the nonlinear system, then using a linear system to Kalman filtering, to solve the state estimation problem of the nonlinear system, but it ignores the model error caused by this linearization process, and it will cause filtering error. Another limitation is that EKF requires noise statistics which are known, but in most nonlinear systems, the noise statistics is unknown,if the wrong noise statistics (mean and variance) are applied to the EKF algorithm, it makes the filtering error increases even so the filter divergence.To address these limitations, researchers have proposed to construct a model of the virtual noise to compensate EKF algorithm. This paper focuses on the basic principles of EKF algorithm based on virtual noise compensation and build the matlab simulation platform to compare the EKF algorithm and EKF algorithm based on the virtual noise compensation in a simulated underwater environment conditions, use a chart to demonstrate the effectiveness, feasibility, advantages of the improved algorithm. Simulation results show that EKF algorithm based on the virtual noise compensation significantly improves the performance of nonlinear filtering, it solves the accuracy and robustness issues of SLAM.