Onboard Detection and Correction of Orbital Registration Errors Using Rover Imagery

Abstract

Orbital imagery is a key component in planning and executing rover missions. It can be used to predict hazardous terrain, target science objectives, or plan traversal paths, yet the utility of these products is crucially tied to their registration accuracy relative to a rover’s location. Misregistrations of orbital data can lead to problems ranging from unsafe navigation to incorrect or misleading science findings. As rovers become more autonomous it is important that techniques are developed to perform onboard reasoning about terrain information in real-time. This paper outlines a multiresolution approach for detecting and correcting for orbital misregistration errors. Using a path generated from a series of navigation camera images, a mapping is made to high-resolution orbital imagery. This high-resolution imagery is then linked to lower-resolution multispectral imagery, and a direct transformation from the rover’s ground location to high-and low-resolution imagery is recovered.