A model of extended BDI agent with autonomous entities (integrating autonomous entities within BDI agent)

Abstract

In this paper a model of an extended BDI (EBDI) agent with autonomous entities and an architecture of the EBDI agent are introduced. The architecture consists of a beliefs module, a desires module, a decision generation module, an integration module, and a group of autonomous entities. In order to represent EBDI agent's mental attitudes (beliefs, desires, and intentions), we define L_m language of an agent's mental attitudes and describe a process of the intention creation. Furthermore, we define L_e language of an EBDI agent's mental attitudes content (spatial relations with temporal constraint and temporal relations). A disadvantage of the BDI model is the inability to adapt to changes in dynamic environment. When an unpredictable situation occurs and there is no applicable rule for that situation, an intention cannot be executed. Therefore, the EBDI model integrates agent with cognitively meaningful internal representation with computationally efficient autonomous entities. The purpose of the integration is to create an agent that possesses mental attitudes but it is adaptive in dynamic environments. When the EBDI agent does not have an applicable rule for current situation, the integration process supported by autonomous entities starts. Autonomous entities create a new representation of the dynamic environment (the new world model). The integration module takes the new world model, finds the solution and creates the new rule. The feasibility of our model has been validated with an example simulated in multiagent programmable modeling environment.