Abstract:In recent research of mental state in MAS, researchers try to add social mental state attributes into individual model, and to study the reasoning rule of social commit, dependence, and joint intention. BDI model is applied in the specification and verification of practice system successfully, such as dMARS. But, there are many shortcomings with respect to the researches on BDI agent, such as the over-domination of the intention-centered methodology, inconvenience to the description of sociality in MAS, and the gap between theory and practice. This paper replaces intention by obligation, which is a social concept and more intuitive. Firstly, we review the research on mental state of agent, analyze the shortcomings of the intention-centered methodology, present the layered conceptual model of MAS. In individual layer, agents memory and process belief, desire, and intention. In social mental layer, agents contact with other agent by memory and process commit, and obligation. The mental state of social layer is related with individual layer of agents. Then we point out that belief, desire and obligation (BDO) should be regarded as primitive attributes to describe agent's mental state and various societal phenomenon. We develop BDO logic and its semantics, import reward, punish, commit and decommit, and give descriptions for such social concepts as team, organization, and organization intention. We also discuss the deductive relationship between obligation and other mental states. We regard BDO logic as an improvement of BDI logic presented by Rao and Georgeff, and as a more expressive framework for the description of social agents and societal concepts. At last, we give an example to show the description power of BDO logic. The future work include giving more perfect semantic model, proving the soundness and completeness of BDO logic, giving dynamic model of BDO agent by combining dynamic revise semantic of mental states, and presenting agent/MAS structure based on BDO logic frame.

Graded BDI models for agent architectures based on Łukasiewicz logic and propositional dynamic logic

An Implementation of BDI Model Based on Rule Reasoning

ANGLE: an Autonomous, Normative and Guidable Agent with Changing Knowledge.

An Argumentation-Based Flexible Agent with Dynamic Rules of Inference

On Truth-Gaps, Bipolar Belief and the Assertability of Vague Propositions

Modeling Social Agents in BDO Logic

Agent-BDI Logic

An Extended Bdi Agent With Policies And Contracts

Modeling belief, capability and promise for cognitive agents – a modal logic approach

Value-based Practical Reasoning: Modal Logic + Argumentation

Agent intelligence model based on dynamic fuzzy logic

A Model of Mental States Combined with Utility for Agents

Argumentation-Based Non-Monotonic Reasoning of Agents

Forming mechanism of BDI agent based on activity and its application

An Extended Agent BDI Model with Norms, Policies and Contracts

PD-Agent: A Flexible Agent BDI Model for Autonomic Computing

Toward a formal theory of belief, capability and promise incorporating temporal aspect

An elementary belief function logic

A Dynamic Epistemic Framework for Reasoning about Conformant Probabilistic Plans

A Model for Learning and Acting Rationally in Multiagent System.

General Dynamic Dynamic Logic.