A prolog meta-interpreter for AgentSpeak(L)

Abstract

Software agents have gained much attention among the research community in the last years. With characteristics as autonomy, reactivity and social ability, software agents are capable of acting independently without requiring supervision or guidance from another higher level entity. There has been developed a whole range of different theoretical models that could be used, such as BDI model, as well as languages that satisfy efficiently the requirements of multi-agent systems. So far, AgentSpeak(L), is considered as the most commonly known approach to a concrete language in agent-programming. Jason, an actively developed interpreter of AgentSpeak(L), is written in Java, seems like a contradiction, since AgentSpeak(L) was introduced as an extension of Logic Programming. In the current project a Prolog meta-interpreter of AgentSpeak(L) is presented, capitalizing on the features of logic programming that seem to be very close to the ideas of agent-oriented programming. Tackling the challenge of implementing a robust and efficient framework for software agents, under which they would be able to act autonomously and communicate with each other, this project aims to combine the features of modern logic programming with the agent-oriented systems. The proposed meta-interpreter stays close to the original syntax of AgentSpeak(L) as well as extends the set of features, implementing new ones, such as strong negation, belief and plan annotations, complicated mental rules and failure handling, as those were proposed in later research in the AgentSpeak(L) model. The communication mechanism is based on the Pedro platform, a subscription/notification system. Based on the server-client model, agents are able to exchange messages directly and in an asynchronous way. The established communication network is managed from an independently implemented entity, the registry, which is responsible for updating the agent list, including all the active agents that take part in the Multi-Agent system.