CISE Postdoc: Exact Topological Localization without Explicit Localization

点击翻译按钮获取中文摘要

9808271 Choset, Howard M. Carnegie Mellon University CISE PostDoc: Exact Topological Localization without Explicit Localization Localization is crucial to the realistic implementation of sensor based exploration, which is a task that enables a robot to explore and build a map of an unknown environment. The robot needs localization to determine if it is exploring new territory or re-encountering an already mapped region. First, the proposed work suggests a new map, termed the topological Voronoi graph, which only stores the nodes, adjacency relationships between the nodes, and the distance between nodes of the generalized Voronoi graph structure. Instead of using complicated sensor and encoder error models to determine the robot's location, the robot exploits the topology and geometry of the topological Voronoi graph to determine its location in this graph, and thus infer its location in the work space. This method is theoretically well founded and simpler to implement than conventional methods. Furthermore, the proposed method extends into any dimensioned space where conventional methods are computationally intractable. The proposed work will be demonstrated in a 10,000 square feet area. The Post-doctoral candidate, having mastered the Voronoi graph approach, will bring his vast experience in mobile robot programming.