Topological-semantic mapping, localization and navigation
ROBOCOMPLEX
ROBOCOMPLEX
Underground environments present a set of unique challenges that make it difficult for traditional navigation technologies to operate in. For this reason, we propose the development of novel and more tailored navigation approaches for these environments. The specific tasks are:
Topological navigation of a UGV in underground environments.
Topological mapping in underground environments.
Mapping from heterogeneous robot teams and sensors.
High-speed traversal of tunnels with UAVs.
Coordination of heterogeneous robot teams under connectivity constraints.
Algorithms for topological-semantic mapping, localization and navigation that allow the autonomous operation of both UGV and UAV robots in underground environments. By focusing on topological-semantic features, it is possible to exploit the underlying structure of subterranean environments.
Two USVs: (Roboboat 1) for autonomous navigation and 3D mapping of partially flooded caves or mines; and Roboboat 2 for water channels inspection in tunnels. Custom metrics have been defined to iteratively validate and improve the 3D model of the passageways.
A general graph-based framework localization technique, seamlessly integrates LiDAR, inertial and GNSS measurements, and cloud-to-map alignment in a sliding window graph fashion, which allows to accommodate the uncertainty of each observation. The modularity of our framework allows it to work with different sensor configuration (LiDAR resolutions, GNSS denial) and environmental conditions (map-less regions, large environments).