Utility and applications of rover science autonomy capabilities: Outcomes from a high-fidelity analogue mission simulation

Abstract

Abstract The 2016 CanMars analogue mission simulated a Mars rover operation modeled very closely on NASA s Mars 2020 mission, with a comparable rover, instrument suite, and operations system. Over 28 sols of subjective mission time, the mission explored a terrestrial Mars analogue site with a remote science operations team, subject to realistic operational constraints, considering Earth-in-the-loop requirements, downlink timing and budgets, on-board resources, and instrument limitations. In that context, the mission also conducted an extensive set of experiments in employing rover science autonomy – capabilities where the rover decides for itself which data to gather, based on guidance from Earth but without complete specification of the targets. Such approaches have begun to be deployed in exploration missions, and can greatly enhance mission science return. These experiments aimed to understand under which operational circumstances, and for which kinds of observations, each of a suite of science autonomy capabilities was useful and practical. These included blind targeting of remote sensing instruments, autonomous geological scene classification and instrument targeting, visual target tracking, conditional sequencing, contingency activity lists, and precise rover return to previous positions. We present here a case study of these experiments and insights into applicability to rover mission planning.