A two-level approach for robot navigation in large-scale, unknown and dynamic environments

In this paper, an approach is presented for exploration, path planning and navigation by an autonomous robot in large-scale, unknown and dynamic environments. A distinction between a local level and a global level is proposed. At a local level, collision-free navigation is realised based on a local map of the immediate environment. At a global level, a global map is built, capturing the topological structure of the environment. This global map is used for path planning and execution of an exploration strategy. Test results, obtained through simulation, are presented to demonstrate the capabilities of the approach