Autonomous Agents with Application to the Evaluation of Organizational Structures

Abstract : Experimental investigation of adaptive command and control (C2) organizations is limited in scope by the availability of qualified subjects and the complexity of experimental design and analysis for large organizational structures. These limitations challenge the study of adaptive architectures for command and control (A2C2) to represent a realistic command and control environment with a small number of human participants. This paper presents a method of representing large organizations by introducing autonomous agents that simulate additional decision-makers. These agents not only interact with the human participants via message communication, but they also interact with the environment, which indirectly affects the human participants and contributes to a more realistic environment. Since the agents act as additional uncontrolled factors and increase the variability of the experiment, it is important to control their actions such that the variability is minimized (or at least controlled). In this paper, the controllability issue is addressed by scripting agents' actions. The paper also identifies some of the challenges involved in the development of truly interactive and collaborative agents for developing, assessing, and training large C2 organizations, and suggests a course of action for the development of such agents.