An environment for distributed collaboration among humans and software agents

C. Martin, D. Schreckenghost, P. Bonasso, D. Kortenkamp, T. Milam, C. Thronesbery
2003 Proceedings of the second international joint conference on Autonomous agents and multiagent systems - AAMAS '03  
processing, mismatches between a human's mental models and the This paper describes an implemented software prototype for the agent's implementation models, lack of adequate adjustability for Distributed Collaboration and Interaction (DCI) system, which the agent's autonomy, lack of notification to the human about helps humans to act as an integrated part of a multi-agent system. important events at appropriate levels of abstraction, and a basic Human interaction with agents who act
more » ... who act autonomously most of the incompatibility between a human's natural interface modalities time, such as a process control agent in a refinery, has received and the usual interfaces provided by an autonomous control agent. little attention compared to human interaction with agents who We have developed the Distributed Collaboration and Interaction provide a direct service to humans, such as information retrieval. (DCI) system to address these difficulties in human-agent This paper describes how liaison agents within the DCI system interaction and to create an environment in which humans and can support human interaction with other agents that are not, by mostly-autonomous software agents together can form an design, human-centric but must be supervised by, or coordinated integrated multi-agent system. The initial motivation for the DCI with, humans. The DCI system provides a step toward future system arose from our experiences with deployed intelligent seamless integration of humans and software agents into a control agents for NASA advanced life support systems. These cohesive multi-agent system. intelligent control agents monitor and perform process control for regenerative life support systems, which recover usable water or Categories and Subject Descriptors air from the waste products created by biological systems over 1.2.11 [Artificial Intelligence]: Distributed Artificial Intelligence time. Over months to years of continuous operation of these -intelligent agents, multiagent systems. agents, we discovered many unaddressed needs for human interaction. These lessons learned are detailed in [4] . Through the DCI project, we have addressed and enhanced our General Terms understanding of these interaction needs and have begun to Design, Human Factors formulate solutions. The DCI architecture, as described in this paper, embodies the software capabilities that we have identified Keywords as requirements for supporting human interaction with mostly-Liaison agents, human-agent interaction, distributed collaboration autonomous software agents over long-term operation. and interaction 2. DCI ARCHITECTURE
doi:10.1145/860722.860796 fatcat:6yrw626ylvemxb6ovohrfi7ksy