World Modelling with Local Uncertainty for Robot Navigation

This paper deals with the problem of managing the uncertainty of an environment where a mobile robot is navigating in. This uncertainty is local to each feature detected in the environment. These features are modelled as fuzzy segments, in which their uncertainty is represented as a trapezoidal fuzzy set. An application of this schema, the localisation problem, is also described. In this case an EKF keeps track of the location of the robot with respect to the fuzzy segments map that is being constructed. Thus, the EKF uses the local information stored within the segments.