An approach to global localization problem using mean shift algorithm

This paper describes a global localization algorithm (GLA) for a mobile robot in indoor environments, based on a particle filter. This algorithm uses data obtained by two kinds of sensors: encoder and scanner laser. Given a map of the environment, where the robot moves, a GLA tries to localize the robot on the map by using its sensor data. The map of the environment is preliminary built mixing laser readings from well-known poses of the robot. The mean shift algorithm (MSA) processes the map and obtains a list of features, which is the synthetic map of the environment used in the GLA. The MSA is also applied for each sampling step in order to calculate the importance factor of the particles. The trials have been performed by using a dynamic simulator of a differential drive robot and the 3Morduc mobile robot.