Multi-Agent Programs and their Logics

4^th October 2011, 14:00 G12
Dr. Mehdi Dastani
Intelligent Systems Group
Utrecht University
NL

Abstract

The development of multi-agent systems requires programming languages that facilitate the implementation of autonomous agents as well as mechanisms that control and regulate agents' behaviours. It also requires computational mechanisms and logics to execute and verify multi-agent programs. In this talk, I give a brief overview of recent developments in this research field and discuss concepts, abstractions and mechanisms that are used in multi-agent programming approaches. In particular, I will focus on our homegrown BDI-based agent programming language 2APL (A Practical Agent Programming Language) and norm-based organisation programming language 2OPL (Organisation Oriented Programming Language). The first programming language is designed to implement individual agents in terms of belief, goals and plans, and the second programming language is designed to implement multi-agent organisations in terms of norms and sanctions. The logics developed to reason about the executions of such multi-agent programs will be presented. Finally, it will be shown how norm-based programs can be related to mechanism design, setting the stage for studying formal properties of norm-based programs such as whether a set of norms implements a specific choice function in a specific equilibria.