The Simultaneous Localization And Mapping For AUV In Large-scale Unknown Underwater Environment

In the unknown complex undersea environment,the majority of sensors which are used in the air are unable to use in the undersea environment,such as optical and wireless,which attenuate too soon in the water. In the circumstances of lacking external navigation support,AUV can now only rely on the self-contained inertial sensors and sonar. Forward-looking sonar and its system as the main sensory (sensing equipment),charged with target location,identification and imaging tasks. It can provide the distance and angle data of the obstacles and distinguish the outline and location of target in the two-dimensional space (XY plane).The advantage of using forward-looking sonar is that the sonar can be equipped in the front-end of the AUV,sonar detection provides the overlapping images frame,this will also help to achieve Simultaneous Localization and Mapping algorithm.Simultaneous Localization and Mapping is a focus in the robotics technological field, and is the key of robots autonomous navigation. The Simultaneous Localization and Mapping algorithm does not require the help of prior environment information map, when AUV navigates at the bottom of the sea, it will measure its bottom's tracking speed,depth, acceleration and attitude angle by using self-carrying Doppler Velocity Log(DVL) and MTi sensor, and scan seabed environment by using mechanically scanning imaging sonar(MSIS), then extract useful information in the environment and locate itself, in the meantime, it will construct an feature map of undersea environment.This paper firstly discusses the principle of the Simultaneous Localization and Mapping algorithm and the convergence property of the environment map which is obtained through SLAM algorithm.In addition ,the paper makes a deep analysis of the algrithom's problem in the application. For the the non-linear property of AUV's state and the observation model, a systemic execution of the SLAM algorithm based on the Extended Kalman filter (EKF) is presented.Aimed at the seabed environment's feature ,the paper proposes a method to deal with sonar data into point feature,which not only solves the problem of seabed environment's anomalous of features, but also overcomes the difficulty of large calculated amount when line feature is used to updates. In order to improve positioning accuracy,once receive a set of sensor's measurements,AUV will carry out sensor's data updation and obtain sensor data's optimal estimation to predict the location of AUV. In order to verify the algorithm's feasibility, the paper carry out virtual simulation and reality testing. The result is that, with respect to pure dead- reckoning, the SLAM algorithm can improve the system's positioning accuracy,and verify the algorithm's feasibility in the application of underwater navigation.