Teamcore Project Control of Agent-Based Systems (COABS) Program

摘要

An increasing number of agent-based systems now operate in complex dynamic environments, such as disaster rescue missions, monitoring/surveillance tasks, enterprise integration, and education/training environments. With this increasing population of available agents, we can expect another powerful trend: the reuse of specialized agents as standardized building blocks for large-scale systems. System designers can integrate these existing agents to construct new multi-agent systems capable of solving problems of greater complexity than those addressed by the individual agents themselves. Integrating agents to perform real-world tasks in a large-scale system remains difficult. As part of DARPA's Control of Agent-Based Systems (CoABS) program, the Teamcore project addressed this challenge of agent integration by focusing on general-purpose teamwork capabilities. Based on successful applications of teamwork to closed multiagent systems, the key hypothesis behind Teamcore is that teamwork among agents can enhance robust execution even among heterogeneous agents in an open environment. No matter how diverse the agents may be, if they act as team members, then we can expect them to act responsibility towards each other, to cover for each other's execution failures, and to exchange key information. This report describes key contributions of the Teamcore project in areas of teamwork theory, team monitoring, adjustable autonomy and team oriented programming.