Research On Data Association For Simultaneous Localization And Mapping

This thesis is about data association for feature-based mobile robot and SLAM. SLAM is the abbreviation of Simultaneous Localization and Mapping. Data association is the most important and difficult problem in the SLAM. First of all, thesis introduces the SLAM academic frame, the most important problem and the history of SLAM. Follow this, thesis analyses the reasons why we make use of the data association technology in the SLAM. The history about the data association research has been presented and the difficulty and importance of data association in SLAM also has been analysed particularly in the thesis.Secondly, thesis implements the feature extraction which is the foundation of the data association for solving the complex data association. The method of extracting line feature has been done and the laser model has been set up to abstract the data of the laser measure. The procedure and arithmetic of the line feature extraction about the laser measurement is developed particularly.Thirdly, many research techniques are introduced and compared in the paper. The algorithms of the nearest neighbour(NN) and Joint compatibility branch and bound(JCBB) are researched mostly. The MatLab experimental result proves that JCBB is accuracy but the time of association computation is too long. To overcome this issue, this thesis deals with the JCBB by linearization and the computation efficiency is better. Thesis extracts the line features with the laser model from the raw observation data, then these features will be filtered by the new data association approach with JCBB. A hypothesis vector between the observation and prediction of the measurement will be created. That is the purpose of the data association procedure. Localization of the robot is updated with the extend Kalman filter after data association accomplished.Fourthly, thesis analyzes the importance and approach about the data association in the globle map location. The emulatinal result proves that the linear JCBB approach makes application of the autonomous robot more feasible and available.Finally, a summarization is produced, and also discuss the future work.