An Agent-based Predictive Engine for Variable Star Observing

We consider the problem of designing and implementing an autonomous intelligent agent that can make the most efficient use of the finite amount of telescope time available to an observer. Such an agent must be able to robustly quantify how well an observing schedule is proceeding and thus make changes, in real time, to optimise such a schedule. We present a system of metrics that will form the predictive algorithmic engine for the eSTAR intelligent agents. We discuss the broader applicability to the field of variable star observation.