Application Of Data Association On Simultaneous Localization And Mapping For An AUV

With the development of ocean technology, autonomous underwater vehicle (AUV) becomes a hotspot in the marine studies. To successfully complete underwater tasks, the ability of autonomous navigation is necessary for an AUV. Simultaneous localization and mapping (SLAM) is one of the most promising ways to achieve autonomous navigation. Prior map information are not needed in the algorithm of SLAM, by using the environmental information provided by sensors an AUV carries, self localization and map construction can be achieved. There are many ways to implement SLAM, and in this paper, SLAM based on extended kalman filter (EKF) is mainly studied.SLAM algorithm consists of two components: state estimation and data association. State estimation is the process of estimating the location of AUV and the position of environmental features. Data association is to find the relationship between two different features which are observed at different time and different places, finally decide whether or not the two features correspond to the same object in the physical surroundings. State estimation and data association are two interactive processes, correct data association is the precondition to precise state estimation. As a result, data association plays an important role in the algorithm of SLAM.This paper gives a detailed description to the theory of data association. Besides, with the help of AUV platform, which is developed by our own laboratory, this paper mainly studies the EKF-based SLAM, putting focus on the individual compatibility nearest neighbour (ICNN) and joint compatibility branch and bound (JCBB) data association algorithm. Both simulation and real experimental data show that the JCBB is much more robust in dealing with data association problem in SLAM. An analysis and improvement of JCBB's computation complexity are also given to realize precise vehicle localization and accurate map construction.