Application of SLAM Algorithm Based on Image Sonar to AUV Integrated Navigation

To improve the long-distance or deep-sea navigation and localization precision for an autonomous underwater vehicles(AUV) within limited volume, power and cost, an algorithm of simultaneous localization and mapping(SLAM) based on image sonar is presented for AUV underwater integrated navigation system. According to the data of real-time distance between each feature on map and an AUV, the data of three-dimensional bearing from image sonar and the data of navigation from strapdown inertial navigation system(SINS), the carrier state and map state can be estimated and measured simultaneously and continuously by extended Kalman filter(EKF). The errors are fed back to SINS for correction, hence the accumulated errors of attitude, velocity and position increasing with running distance and time can be effectively inhibited. Furthermore, the accuracy of features identification can be improved by putting acoustics size element into map features vector. Simulation result shows that this method can effectively inhibit errors of inertial navigation by continuously measuring valid features on map. Particularly, the localization accuracy of an underwater integrated navigation is significantly enhanced when same landmarks are observed repeatedly in cruise or return voyage of an AUV.