Intelligent Software Agents for Future Space Explorers: A Conceptual Framework

摘要

We present a conceptual framework for understanding characteristics and interactions of spacecraft/instrument agents in remote environments where success will depend on acquiring and completing analysis for selected targets with minimal human intervention. Optimal ISA modality/architecture depends on the nature of activity, ranging from data-based, sensing, inductive mode for data taking, to model-based, intuitive, deductive mode for pattern recognition. Optimal modality combinations are most critical at the tactical level, where coordination of operations and input to strategic input assessment and planning occur. We develop a software simulation scheme for determining the ordering of these modalities as a function of task.