Terrain roughness identification for high-speed UGVs

Abstract

High-speed navigation of autonomous Unmanned Ground Vehicles (UGVs) in rough unknown terrains requires the detection and identification of the terrain in order to make effective navigation decisions. This paper investigates a geometrical approach to identifying terrain based on its roughness using the terrain elevations from a point cloud generated using a 3D camera. This roughness, called the Roughness Index (RI), is used to identify different terrains by overlaying the terrain with a grid map and using the standard deviation of the point cloud elevations in each grid cell. The experimental testing and results of this terrain identification technique are presented as determined from field experiments using an experimental UGV test platform on rough outdoor terrains.