Models of Interaction

In multi-agent systems, an agent needs to reason about its beliefs, goals, desires, intentions, and those of others, in order to make informed and context-dependent decisions about what actions to take. Moreover, when an agent is situated in a multi-agent "society", it should incorporate notions such as obligations and social norms into its reasoning. Our work involves developing new models and techniques for capturing and specifying the interactions that take place between cooperative or competitive agents. The group conducts research on issues such as teamwork and collaboration, social and organisational issues, negotiation strategies, negotiation with incomplete and dynamic preferences, and interaction protocols. Past projects have included development of a teamwork framework for robots competing in the RoboCup Soccer competition.

The group interacts with a number of industry partners through collaborative research projects, including Agent Oriented Software Pty. Ltd., the Defence Science and Technology Organisation (DSTO), and Neuragenix Pty. Ltd.

Contact: Liz Sonenberg

Publications: Models of interaction