An examination of different coordination techniques in a semi-adversarial system

This research examined artificially intelligent agents that cooperated towards a common goal in a semi-adversarial environment. In a semi-adversarial environment, one group of agents works towards a particular goal that in some way involves another group of agents who neither aid nor hinder the coordinated group. In this research the goal was for the intelligent agents to move the agents of the other group to a specified location in the environment. This study dealt with the artificial intelligence and communications necessary for the intelligent agents to achieve this goal. The agents had to react to the changes in the environment caused by the sporadic movements of the individuals within the unorganized group and also had to coordinate with one another. This research studied three levels of coordination: accidental, implicit, and a more informed implicit coordination. These levels pertained to the amount of information that the robotic agents were programmed to communicate to one another. The amount of information communicated ranged from no communication at all in accidental coordination to all available visual information being communicated in the more informed version of implicit coordination.