Generation of probabilistic graphs for path planning from stochastic maps

This paper presents a method to automatically generate probabilistic graphs for path planning out of stochastic maps. A sample based technique is used to generate a set of paths from which a graph structure of the map is built. The nodes of the graph represent places in the map and the edges the paths between those places, which are labelled with the cost of traversing the edge and the probability of being navigable. We provide results of the proposed method using different types of stochastic maps such as feature-based maps, occupancy grid maps and pose graph maps.