Thispaperdiscussesa proof-of-conceptprototypefor ground-basedautomaticgenerationof validatedrover commandsequences from high-level science and engineeringactivities.Thisprototypeis basedonASPEN, the AutomatedSchedulingand PlanningEnvironment. ThisArtificial Intelligence(AI)basedplanningand schedulingsystemwillautomaticallygeneratea command sequencethatwillexecutewithinresourceconstraintsand satisfy flight rules. Commandingthe roverto achieve missiongoalsrequiressignificantknowledgeof therover design,accessto the low-levelrovercommandset, andan understandingof the performancemetricsrating the desirabilityof alternativesequences.It also requires coordinationwithexternaleventssuchas orbiterpasses and day/night cycles. Anautomatedplanningand schedulingsystemencodesthis knowledgeand uses searchandreasoningtechniquesto automaticallygenerate low-levelcommandsequenceswhilerespecting rover operability constraints, science and engineering preferences,and also adhering to hard temporal constrain...
Rob Sherwood, Andrew Mishkin, Tara A. Estlin, Stev