摘要: |
Unmanned ground vehicles (UGVs) must rapidly and robustly characterize the nature of the terrain they are traversing, to improve autonomous mobility. This research program has focused on the development of a framework for self-supervised terrain classification, which allows a UGV to automatically learn the properties of terrain without human guidance. Work has also focused on novel applications of the self-supervised terrain learning approach, including urban/semi-urban driving on road networks. Finally, research has led to the development of novel sensing techniques for analyzing robot-terrain interaction mechanics at the micro scale. |