Artificial intelligence for real-time monitoring and control

The author describes MARVEL, an automated monitoring and analysis environment with multiple expert systems that has been developed and successfully applied to the automation of interplanetary spacecraft operations at NASA´s Jet Propulsion Laboratory. Automated monitoring and knowledge-based diagnostics of spacecraft anomalies have been part of MARVEL since it was used to monitor the Voyager mission´s Neptune encounter in August of 1989. MARVEL has subsequently been expanded to include the results of research on cooperating expert systems, distributed computing, and real-time metareasoning. The author surveys the generally applicable approaches and implementations that have been developed in each of these areas and describes operational capabilities and results that have been obtained