Continuous mapping and localization for autonomous navigation in rough terrain using a 3D laser scanner

Abstract

For autonomous navigation in difficult terrain, such as degraded environments in disaster response scenarios, robots are required to create a map of an unknown environment and to localize within this map. In this paper, we describe our approach to simultaneous localization and mapping that is based on the measurements of a 3D laser-range finder. We aggregate laser-range measurements by registering sparse 3D scans with a local multiresolution surfel map that has high resolution in the vicinity of the robot and coarser resolutions with increasing distance, which corresponds well to measurement density and accuracy of our sensor. By modeling measurements by surface elements, our approach allows for efficient and accurate registration and leverages online mapping and localization. The incrementally built local dense 3D maps of nearby key poses are registered against each other. Graph optimization yields a globally consistent dense 3D map of the environment. Continuous registration of local maps with the global map allows for tracking the 6D robot pose in real time. We assess the drivability of the terrain by analyzing height differences in an allocentric height map and plan cost-optimal paths. The system has been successfully demonstrated during the DARPA Robotics Challenge and the DLR SpaceBot Camp. In experiments, we evaluate accuracy and efficiency of our approach.