CISE Postdoc: Exact Topological Localization without Explicit Localization

9808271 Choset，霍华德M. 卡内基梅隆大学 CISE PostDoc：精确拓扑本地化，无需显式本地化 定位是基于传感器的探索，这是一个任务，使机器人探索和建立一个未知的环境地图的现实实现的关键。机器人需要定位，以确定它是在探索新的领域还是重新遇到已经绘制的区域。 首先，所提出的工作提出了一种新的地图，称为拓扑Voronoi图，它只存储节点，节点之间的邻接关系，和广义Voronoi图结构的节点之间的距离。代替使用复杂的传感器和编码器误差模型来确定机器人的位置，机器人利用拓扑Voronoi图的拓扑和几何来确定其在该图中的位置，从而推断其在工作空间中的位置。 该方法在理论上是有根据的，并且比传统方法更容易实现。此外，所提出的方法扩展到任何维的空间，传统的方法是计算困难的。 拟议的工作将在10,000平方英尺的区域进行演示。博士后候选人，掌握了Voronoi图的方法，将带来他在移动的机器人编程的丰富经验。

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.