Affiliations: [a] School of Electrical and Information Engineering, Knowledge Based Intelligent Engineering Systems Centre, University of South Australia, Mawson Lakes SA 5095, Australia | [b] School of Computer and Information Science, University of South Australia, Mawson Lakes SA, 5095, Australia
Abstract: Multi-agent teaming is an active research field of multi-agent systems. Flexible multi-agent decision making requires effective reaction and adaption to dynamic changes under time pressure, especially in real-time and dynamic systems. The joint intension and sharedplan are two most popular theories for the teamwork of multi-agent systems. However, there is no clear guidance for designing and implementing agents' teaming. BDI (Belief, Desire, and Intension) architecture has been widely used to design multi-agent systems. In this paper, a role-based BDI framework is presented to facilitate the team level optimization problems such as competitive, cooperation and coordination problems. This BDI framework is extended on the commercial agent software development environment known as JACK Teams. The layered architecture has been used to group the agents' competitive and cooperative behaviors. In addition, we present the reinforcement learning techniques to learn different behaviors through experience. These issues have been specified and investigated within a real-time 2D simulation environment known as soccerBots.
Keywords: Agents, BDI architecture, cognitive architecture, JACK agent platform, role, multi-agent teaming, decision making