The traversal of rough a priori unknown terrain by an Unmanned Ground Vehicle (UGV) at high-speed requires the assessment of terrain in front of the vehicle as well as algorithms to select a speed based on the identified terrain. Current techniques are either susceptible to miss-classification, do not address the challenge of speed selection, or have not been experimentally tested.

This thesis proposes a geometric terrain identification technique where a range sensor captures a 3D terrain point cloud and uses the standard deviation of the terrain point elevations to determine a roughness score. Using this roughness score three methods were proposed in this thesis to select a fast but safe allowable speed for a UGV based on predicting the force the terrain will exert on the UGV while driving. These methods were experimentally tested on rough outdoor terrain to demonstrate and compare their performance.