Research On Data Association For Auv Simultaneous Localization And Mapping

Simultaneous Localization and Mapping (SLAM),as an extremely challenging topic in robot navigation field, has been widely studied and made a significant progress. SLAM does not require prior information of the underwater environment. When moving underwater, the mobile robot uses its on-board sensors to perceive the environment and extract the useful information to localize itself, meanwhile, incrementally build a map of the underwater environment. In the complex undersea environment, most of the external sensors are not available, Autonomous Underwater Vehicle (AUV) can only rely on the on-board inertial sensors to obtain its own information and get the environment information by the active sonar. However, the observation information obtained by the sonar does have serious noise interference which causes great difficulties to the navigation and localization of AUV and puts forward higher requirements to the SLAM algorithm.In the SLAM algorithm, data association is one of the most crucial and difficult problem. Especially in the complex underwater environment, reliable data association is particularly important. It is the process of determining the correspondence between features observed in the environment and features viewed previously or existed in a map. Correct data association will allow the robot determine its own position relative to the features observed and just few times of association failure may lead to diverge of the SLAM algorithm.In this paper, the theory of data association is presented, the framework and specific implementation steps of EKF SLAM based on AUV are described and then the key step—data association is highlighted. Simulation experiments are carried out to compare the two prevailing data association algorithms—Individual Compatibility Nearest Neighbor (ICNN) and Joint Compatibility Branch and Bound (JCBB). The simulation results show that it is more robust for JCBB algorithm to deal with the data association problem of SLAM, however, the disadvantages of JCBB are large computational complexity and slow computation speed. So some improvements on the computational complexity of JCBB are presented in this paper. Through the analysis of simulation results, it can be seen that using the improved data association method in EKF SLAM can significantly improve the robustness of AUV Simultaneous Localization and Mapping.