Strategies of Cooperation in Distributed Problem SolvingDistributed artificial intelligence is concerned with problem solving that is done by groups of agents. This Note describes strategies of cooperation that groups require to solve shared tasks effectively. We discuss such strategies first in a domain-independent fashion, and then in the context of a specific group problem-solving application: collision avoidance in air traffic control. We begin by contrasting the methodologies, difficulties, and opportunities of distributed and centralized problem solving. From this analysis, we infer a set of requirements on the information-gathering and organizational policies of group problem-solving agents. We then discuss a set of distributed problem solvers that we have developed in the domain of air traffic control and describe some experimental findings with the cooperative strategies used. In particular, we note large task-dependent differences in processing times, communication loads, and system errors between the several cooperative strategies. (Author). |
Common terms and phrases
agent problem-solving situation agent's plan Agents typically air traffic control airspace appropriate agent assigned tasks available expertise communication channel Communication load complete role assignment computation conflict task constraint factor cooperative strategies decomposed difficulties distributed agents Distributed artificial intelligence distributed fashion distributed problem solvers DISTRIBUTED PROBLEM SOLVING domain-independent fashion Execute plan four systems given task globally coherent group problem groups of agents implemented INDIANA UNIVERSITY infix INFORMATION-DISTRIBUTION POLICIES information-gathering and organizational intentions knowledgeable policy least constrained policy least spatially constrained limited knowledge limited-bandwidth channels mean number negative interactions organizational policies plan-fix planner policy of task problem solving Prototypical task sequence scenarios selected agent Send unrelated Separation errors sequence of tasks shared conflict shared convention policy shared resources shared tasks effectively single agent single-agent problem solvers solutions solve problems solve shared tasks strategies of cooperation subtasks task centralization policies task sharing task-sharing policy tasks and communications three task-centralization unrelated activities world model