A hybrid procedural/deductive executive for autonomous spacecraft

Abstract

Tha Ncvr Millennium Remote Agent (NMRA) will be the first AI system to control an actual spacecraft. The spacecraft domain places a strong premium on autonomy and requires dynamic recoveries and robust concurrent execution, all in the presence of tight real-time deadlines, changing goals, scarce resource constraints, and a wide variety of possiblo failures. To achieve this level of execution robustness, wo 11ixw integrated a procedural executive based on generic proccdurca with a deductive model-based executive. A procedural executive provides sophisticated control constructs such as loops, parallel ectivity, locks, and synchronization which arc used for robust schedule execution, hierarchical task decomposition, and routine configuration management. A dcductivc cxccutive provides algorithms for sophisticated state infcrencc and optimal failure recovery planning. The intograted executive enables designers to code knowledge via a combination of procedures and declarative models, yielding a rich modeling capability suitable to the challenges of ranl spacecraft control, The interface between the two exccutivcs ensures both that recovery sequences are smoothly merged into high-level schedule execution and that a high dcgrcc of reactivity is retained to effectively handle sdditional failures during recovery.