Jan 1, 1984
Groups of agents (human or machine) can solve shared tasks effectively by applying cooperative intelligence. Cooperative behavior is necessary for solving problems that, because of time or other physical constraints, cannot be solved by one agent acting alone. Complex, spatially distributed military systems — such as tactical air operations; naval task force control; and command, control, communications and intelligence networks — frequently rely on cooperative problem solving. This report develops aspects of cooperative intelligence in the context of a specific application, coordination of groups of remotely piloted vehicles in a surveillance mission. The findings suggest that (1) a combination of object-oriented simulation and logic programming appears to provide an effective framework for exploring and implementing distributed problem-solving systems, and (2) choice of task negotiation procedure, message passing protocol, planning algorithm, and uncertainty representation technique depends strongly on situational conditions such as time stress, communication costs, and number of planning options.