SLAM with conceptualisation

In this paper, we presented a novel approach for robot Simultaneous Localization and Mapping (SLAM) in an indoor environment. We observed that such an environment is partitioned into rooms and corridors. We thus propose that the robot should compute a map beginning with a conceptual view of its environment rather than a physical view of it. Our approach thus simplifies the localization problem by not having to consider all surfaces perceived. This is especially useful since in the real world, the environment is often cluttered with objects. Instead, and as shown in our experiments, the SLAM problem is solved by juxtaposing representations of rooms and corridors computed. Our approach was tested using a mobile robot equipped with an odometer and a laser range finder. Our initial result shows our robot successfully juxtaposed three rooms to form a correct global map.