Track:
Contents
Downloads:
Abstract:
In this paper we present some preliminary formulations of temporal approximations and refinements. These are respectively the processes of abstracting away details about temporal objects, and the processes of adding more detail to them. Intelligent robots operating in the world will not only need to reason temporally, but they will need to be able to approximate or refine their reasoning as necessary in order to operate efficiently. The four kinds of refinements/approximations we study are treating ramifications as internal events, already proposed in (McCarthy, 2002), elaborations of narratives, expansions of events, and increasing the predictive capacity of theories.