PAULA: Multi-agent architecture for coordination of intelligent agent systems

This paper describes an architecture named PAULA that enables coordination of autonomous agents by means of knowledge based on the agents´ perception about cooperative environments. Autonomous agents are then classified into supervisor agents and physical agents. In addition, a method to improve cooperative task performance has been designed using the electronic Institutions (e-I) features. In this sense, the supervisor agents must coordinate among themselves to determine which order to perform the actions as must the physical agents to perform task allocation process, thereby improving the cooperative task performance through suitable decision making. Finally, this paper illustrates the effectiveness of our approach by presenting a solution to cooperative actions on a robot soccer testbed.