ROS 2 Integrated Continuous Terrain Mapping Comparison for Autonomous Navigation in Unstructured Environments

摘要

The space exploration with robotics and autonomous systems presents significant technological challenges, particularly in navigation. As interest grows in planetary exploration, the need for robust autonomous navigation systems becomes increasingly urgent. This paper addresses the critical component of mapping in autonomous navigation of space rovers. We explore two distinct mapping pipelines for unstructured outdoor environments, comparing their methodologies and performance in generating traversability maps with slope information. Although both approaches share similar input and output formats, they diverge in their computational techniques, with one relying on surface normals computation and the other on Digital Elevation Model (DEM) calculation. The comparison is carried out within the ROS 2 environment, encompassing both static and dynamic scenarios. In the static context, both pipelines receive a detailed point cloud of a facility featuring Mars-like terrain. Conversely, in the dynamic scenario, data are captured by a freely moving stereo camera within the same outdoor environment.