| The situation calculus is a second-order logic language used to describe the characteristics of autonomous agents acting in a dynamic system. Its breadth and powerfulness has been shown by the tremendous work achieved in broad areas. Recently, researchers have become more and more interested in modeling and controlling the performance of agents in an uncertainty system with such language.; In this paper, we focus on the problems that the autonomous agent performs similar strategies repeatedly under same local situations in some uncertainty system. We introduce a special concept of action—macro-action—into the situation calculus, extend basic action theories and regression operators, and develop a knowledge base for the macro-actions so that the agent can remember certain information of them and later “recall” it when the agent performs the macro-actions in the same local situation again, therefore achieving the goal of saving computational time. |
