One of the most important properties that a robot must have in order to be considered autonomous is the ability to localize by itself in an unknown environment, using the information gathered by its sensors. The system uses a cheap web camera, carried by a mobile robot or by a person, while it builds a map and at the same time estimates its localization with respect to this map. To develop such system, problems such as camera calibration, corners detection, features parametrization, motion and measurement models, and the estimation of the joint state camera-features using the extended Kalman Filter, must be tackled.