Overhead Detection, Identification, and Tracking of Multiple Surface-based Exploration Vehicles

As NASA and other space agencies venture out to explore planetary bodies of high interest (Mars, Titan, Europa, Enceladus, etc.), especially from an astrobiological point of view, i.e., the quest for extant/extinct life beyond Earth, planetary field geologists will have to be replaced and emulated by robotic spacecraft, at least for the foreseeable future. As such, these robotic explorers will have to be equipped with observation, analysis, and reasoning capabilities of a field geologist. Moreover, to mimic the geologic approach of local to regional to global reconnaissance in an integrated, mutually informing fashion, these robotic explorers will likely have to operate as part of multi-tiered robotic mission architectures. Several precursors to such mission architectures have been proposed, such as the introduction of an overhead perspective either through a balloon, blimp, airship, or helicopter/rotorcraft. Using an overhead perspective provides many advantages for exploration and reconnaissance, as well as for guidance, navigation, and control (GNC). A real-world instantiation of an overhead perspective is the use of the HiRISE camera aboard Mars Reconnaissance Orbiter for GNC support of the Mars Exploration Rovers. In this context, this paper focuses in particular on the challenge of detection, identification, and tracking of multiple deployed ground-agents, such as rovers on Mars or lake landers on Titan. The devised framework comprises the use of distinct, linearly independent, scaling invariant, and directional templates that are matched to similar markings on top of the respective deployed ground-agents through rotation, transformation, and scaling operations. This allows the spatial detection and identification of the respective ground-agents. The centroids of the detected templates are subsequently tracked simultaneously through the repeated use of this template-matching procedure. This detection, identification, and tracking framework enables the GNC of multiple, and thus expendable, ground-agents from an overhead perspective(s), e.g., as part of multi-tiered exploration mission architectures to access high(er)-risk, but high(er)science payoff regions.