Employing human knowledge to solve integrated coordination problems

An Integrated Coordination Problem involves solving multiple related subproblems that collectively satisfy the requirements of a user, including subproblems that depend on the user´s participation to solve. Fundamental challenges in solving such a problem include defining mechanisms to solve the individual subproblems, formulating the information and control flow between these mechanisms that supports flexible end-to-end problem-solving, and providing access for people to oversee and participate in the problem-solving process. In this paper, we describe a multi-agent architecture that addresses these challenges by embodying mechanisms in computational agents and by treating the collective problem-solving across agents and people as a collaborative process. We argue that our approach exploits concepts that straddle the boundary between collaborative technologies and multi-agent systems, and demonstrate its advantages and capabilities in the context of an emergency medical response scenario.