Planning as Temporal Reasoning.

摘要

This paper describes a reasoning system based on a temporal logic that can solve planning problems along the lines of traditional planning systems. Because it is cast as inference in a general representation, however, the ranges of problems that can be described is considerably greater than in traditional planning systems. In addition, other modes of plan reasoning, such as plan recognition or plan monitoring, can be formalized within the same framework. 1 INTRODUCTION There is strong interest currently in designing planning systems that can reason about realistic worlds. In moving from the toy-world domains that characterized early work, researchers are looking at a wide range of issues, including reasoning in uncertain worlds, interacting with processes and events beyond the agent's direct control, and controlling mechanisms in real-time (i.e. robotics). One of the problems faced in extending existing frameworks is the weak expressiveness of the representation of the acti...