This paper presents an approach for path following in a partially known indoor environment. The approach combines the artificial potential fields technique, for navigation, with a local supervisor. The supervisor uses, a single geometric model and a known path. For each node belonging to the path, the supervisor generates multiple partial goals, including the next planned path point and the mission goal. Then, the supervisor selects, the most adequate, depending on, the possibility of being reached. Some criteria such as the distance travel optimizing, the uncertainty estimation and the localization zones, are considered. The supervisor is built to support the navigation system, which increases the missions success possibilities. The dynamic sub-goals generation and the selection criteria, are proposed. A kinematic model and the uncertainty representation for the robot, are developed. The artificial potential field concept, is introduced. Finally, the experimental results and the conclusions, are shown sor to support the navigation system, is proposed. In second place, the Kinematic model, the uncertainty dealing, are developed. In third place, the artificial potential field concept, is introduced. Finally, result and conclusion, are shown.
展开▼
