Topographical Landmarks for Ground-Level Terrain Relative Navigation on Mars

Many of the tasks planned for future generation Mars rovers rely heavily on having accurate knowledge of the rover's location in a Martian body-fixed coordinate system. Current solutions for localization require regular human intervention in order to detect and rectify drift, and thus stand to benefit from systems that can run in real-time on the rover itself. We study the feasibility and performance of an automated approach to localization in which the rover makes bearing-only measurements to geographic features in its surroundings (hills, boulders, peaked ridge-lines, etc.). When the location of these landmarks can be cross-referenced with a map of Mars, the resulting solution will be globally registered and will help correct any drift during visual-odometry-aided drives. This paper studies two related problems: First, how can we locate geographic features that the rover can feasibly see, when provided an elevation map of the surrounding terrain? We provide a software tool that can extract features from elevation maps for comparison to imagery. However, once these landmarks are identified, it is not obvious if a given path for the rover will contain sufficient features to navigate autonomously. Accuracy will depend on the quantity, range, and relative geometry of landmarks that are available. Thus, the second contribution is to provide a GIS plugin that analyzes the terrain informativeness of large operating areas as well as more specific paths. We present an analysis of Jezero Crater in which Perseverance's Navcam is used as the hypothetical sensor. In certain favorable regions, worst-case localization accuracy in the 10 meter range is achieved from a single set of measurements (comparable to GPS on Earth). The map overlays generated by this analysis have the potential to aid in long-term mission planning by highlighting broad areas of high or low informativeness. These tools are computationally efficient and will be made open source to allow Mars mission planners, formulation studies, and rover drivers to plan for any future image-based self-localization capability.