The researchers propose a new navigation framework called STAGE, which utilizes a two-layered graph representation of the environment for large-scale exploration. This framework also incorporates a new uncertainty awareness scheme to handle dynamic scene changes in previously explored areas. The STAGE planner is efficient in handling changes in the environment, such as blocked pathways, by adaptively updating the obstructed part of the global graph from traversable to not-traversable. The effectiveness of the method is demonstrated through simulation runs and real-world scenarios involving a legged robot equipped with a camera and lidar sensor.
Publication date: 6 Feb 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2402.02566
