In this study, we analyze the problem of detection of depressions or drop-offs in the automated guidance of roving robots. The proposed approach is based on the principle that if one is too near a depression, one is bound to see new information which initially was occluded. To exploit this principle, two steps are undertaken. The first step involves the derivation of the correspondence process to allow the vision system to relate a location of interest in a sequence of frames. The second step involves the development of methods to detect and identify, in this location of interest, the occluded information.
