LiAS: A reflexive navigation architecture for an intelligent mobile robot system

In this paper, a complete hierarchical navigation architecture for applications in real industrial factory environments is presented. An off-line global geometrical planner uses a rough CAD-model of the working environment to plan consecutive via-points which the robot must follow. The navigation between the via-points is performed by a two-level on-line navigation algorithm. It consists of an on-line planning module combined with a low-level fuzzy logic avoidance behavior which enables the robot to move to the nest goal by only specifying its coordinates in even completely a priori unknown and unstructured factory environments, The system includes a docking motion, based on a dynamic guidance technique. A perception fusion module combines information of three different sensors for accurate modeling of the world. The presented navigation method was tested a with the mobile robot Leuven Intelligent Autonomous System (LiAS) and it proved to be useful in real world applications.