In recent decades research and development in the field of robotics has been rapidly growing. A result of this growth can be seen in the wide spread increase of the use of robots in industry. However, the field of robotics does not only consist of robots for industrial factory automation. A lot of research has also gone into service, security and military robots. Many of the applications that fall into the before mentioned categories are mobile robots. For mobile robots to be able to operate safely within their environment they need to be able to estimate their position within that environment. As most environments are dynamic, at least to some extent, a localization system has to be able to cope with that. The system presented in this thesis uses the Adaptive Monte Carlo Localization along with map maintenance method and outlier rejection for localization in dynamic environments. The map maintenance method utilizes so called meta-rooms, re-constructed static structure of rooms, to maintain a map for localization, as it is more reliable to localize against static objects than dynamic. The outlier rejection method filters out laser scan measurements that do not arise from static structures in the environment, these measurements are then used by the localization system along with the maintained map. Experiments were performed on the system in a simulator. In these experiments, the proposed system was compared with the Adaptive Monte Carlo localization without map maintenance or outlier rejection. The results were positive and show that the new system gives less errors than the one it builds upon.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Robot Localization in Dynamic Environments, Robotlokalisering i dynamiska miljöer


    Beteiligte:

    Erscheinungsdatum :

    2015-01-01


    Medientyp :

    Hochschulschrift


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Klassifikation :

    DDC:    629