Anticipatory agents based on anticipatory reasoning

Abstract

Anticipatory agents are agents, which are capable of somehow make predictions in advance based on their own knowledge at present and use the predictions about future to guide their current actions. To construct such an agent, we propose an approach based on temporal relevant logic. We analyze the requirements for logic systems to underlie anticipatory reasoning at first. Secondly, we present a logic family, named "temporal relevant logic," which can satisfy these requirements. And then we propose a conceptual architecture based on anticipatory reasoning in the framework of temporal relevant logic. At last, we present solutions to some issues about automation of anticipatory reasoning. © Springer-Verlag Berlin Heidelberg 2005.