A Logic for Rational Agents Which Handles Beliefs and State Transitions Probabilistically

DOI

Bibliographic Information

Other Title
  • 信念と状態遷移を確率的に扱う合理的エージェント向け論理

Description

In BDI model, Agents' behaviors can be formally described using the temporal logic system named BDI logic. We formerly proposed TOMATOes, as an extension to BDI logic, by introducing probabilistic state transition operator which is necessary for representing some extensions to BDI model such as integration with reinforcement learning. However, TOMATOes does not have probabilistic belief modal operator, which would be useful in some applications. Therefore, it is desired that both beliefs and state transitions can be handled in a single logic. In addition, to be used together with BDI model, such logic should also be able to handle mental state consistencies, which are required properties of rational agents. As a logic which meets these requirements, we present B-TOMATOes, a further extension to TOMATOes, and show some examples of formalizing behaviors of rational agents using probabilistic beliefs.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top