A HYBRID MAPPING APPROACH FOR MOBILE ROBOTS

摘要

In this paper we describe a mapping strategy for a mobile robot. The environment is represented by a graph relating locations of particular places. The qualitative nature of the movements the robot executed to reach such places is also represented in the graph. The sensorial capacity is not demanding; a few distance sensors and a pair of odometers are enough for the robot the build the map from zero. However, some a priori knowledge about the environment is necessary: the fuzzy predicates necessary to classify movements are pre-established, and it is assumed that the robot has a home which innately recognizes. Some examples over a simulated Khepera are presented that validate the model, emphasizing its details and presenting empirical analysis about important topics as are re-localization. The main features of the approach are also discussed in relation to other research efforts.