Autors: Hensel, S., Marinov, M. B., Kupitz, C., Trendafilov, D.
Title: Evaluation of Kalman Filter Configurations for Robot Localization Using Sensor Data Fusion
Keywords: Robot Operating System, Extended Kalman filter, Unscented Ka

Abstract: In this work, three different configurations of the Kalman filter developed by Tom Moore for the Robot Operating System are presented. These form the basis for localization using sensor fusion in the ROS framework used. The aim of this work is the construction and verification of localization for a mobile robot system Husky A200 from Clearpath Robotics. For this purpose, the possibilities of the existing system were examined, and several versions of localization filters were configured. In the end, a verification of the results in different scenarios is compared. For this purpose, the results of a variant of the Extended Kalman filter in 2D (EKF2D), a variant of the Unscented Kalman filter in 2D (UKF2D), and a variant of the Extended Kalman filter in 3D (EKF3D) are verified and compared. The investigations showed that the EKF2D provides the best and most robust results for localization, despite having a 17.3% higher end position deviation compared to the UKF2D variant. The EKF3D confi

References

    Issue

    FDIBA Conference, vol. 6, 2022, Bulgaria, ISBN 2535–132X

    Full text of the publication

    Вид: публикация в международен форум