Autonomous Identification of Locations for Unmanned Ground Vehicle (UGV) Cover and Concealment With the Use of Video Images

摘要

Autonomously identifying appropriate locations that provide cover and concealment for an unmanned ground vehicle (UGV) is a difficult task. This report expands an earlier algorithm for accomplishing this task, based on video images to address the difficulty of handling thin or narrow objects in the UGV environment. The vertical line approach used in the earlier algorithm is expanded to include an analysis of local regions to assess color homogeneity and size criteria. Careful selection of seed pixels attempts to ensure that they represent a potential thin or narrow object and not the background. The expanded algorithm provided very good results, based on the video sequences analyzed.