We have presented a quantitative, feature-based approach for pose estimation from outdoor visual information. We detailed two key steps in the approach: automatic detection of mountain peak features from images, and estimation of image/map correspondences and pose. We presented results with real data which demonstrate an improvement over the state of the art.
Future work will concentrate on two areas. First, we will implement our position estimation approach on hardware that is capable of handling a stream of images arriving at near real-time rates. Second, we will integrate our approach into an augmented reality system that overlays the input image stream with graphical icons on selected terrain features, thus offloading some of the navigation functions from the teleoperator. With these two achievements, we will have made it easier to remotely drive a wide variety of vehicles, and contributed to our return to the Moon.