Mobile Robot Localization Using Fuzzy Maps

This paper deals with the problems of map building and mobile robot localization. Ultrasonic sensors information, obtained as the robot moves, is integrated in order to build a map of the environment. This map is afterwards used for robot localization, correcting the errors that the dead reckoning system accumulates in long displacements. The approach focuses on the way to analyze sensor information and on the way to reconstruct the outline of the objects. Since measuring conditions are unknown and just a small number of observations are normally available to reconstruct the boundaries of the objects, this will generate uncertainty on their real location. Fuzzy sets are used to represent this uncertainty, their degrees of membership indicating the extent to which one boundary can be considered similar to other boundaries located in the proximity. Experimental results with a mobile robot in an office environment are also presented.