AgentSpeak(PL): A New Programming Language for BDI Agents with Integrated Bayesian Network Model

Within the current BDI paradigm of agent oriented programming, it is not possible to develop a practical and straightforward software when agent´s beliefs are based on probabilistic knowledge related to their environment, because the logical base of the programming language do not allow this possibility. Usually, to develop this kind of hybrid agent, it is necessary to make use of advanced programming techniques, which combine and integrate different development environments and programming languages in order to represent the logical part and the probabilistic part of the model, to make implementation feasible. This work presents a new agentoriented programming language called AgentSpeak(PL), based in AgentSpeak(L), a classical BDI programming language. AgentSpeak(PL) integrates the concept of probabilistic beliefs through the use of Bayesian Networks, to core BDI programming concepts. The language is implemented through an extension of the Jason programming environment.