Automated Recovery Planning Assistance for Unknown Anomaly Resolution

The identification and resolution of unknown satellite anomalies currently demands a significant portion of costly technical resources. We have done work demonstrating the feasibility of employing a synergistic combination of model-based and case-based reasoning methods to the anomaly identification task. We now propose to apply this technology to the recovery aspect of the problem. We propose an automatic, intelligent planning and scheduling assistant for recovery planning. By innovatively combining Artificial Intelligence (AI) techniques that we have successfully used for NASA KSC's high-level manifest planning, JSC's spacecraft activity planning and MSFC's payload activity scheduling, we will produce a planning system with capabilities well beyond current methods.We will use AI knowledge representations including objects, semantic networks and rules to capture plan components, constraints and heuristics gathered from expert planners. We will automate recovery planning through the use of the intelligent entities concept, similarity-based and constraint satisfaction techniques. The Air Force will benefit from rapid creation of plans for saving the satellite and for recovering lost function, and the capture of expert planner knowledge. Phase I research and prototype development will lay the groundwork for the Phase II implementation of an automatic planning and scheduling assistant and eventual commercialization of the system.