In many task-planning domains, dynamic assemblies of autonomous agents are replacing hierarchical organisations because they promise more agility. In such assemblies, interdependent tasks might be given to different agents that each make a plan for their set of tasks. The feasibility of a joint plan for the total set of tasks, however, is likely to be endangered: Autonomous planning behaviour might result in individually constructed plans that are not jointly feasible. Therefore, (plan) coordination mechanisms have to be introduced to guarantee that even if each individual agent plans its part of the tasks independently from the others, the result will be a feasible joint plan for the complete set of tasks. In previous work we have addressed a coordination mechanism for moderately-coupled tasks, i.e., tasks that are partially ordered by some precedence relation expressing that some task t has to be completed before another task t can be started. Often, however, we have to deal with m...
J. Renze Steenhuisen, Cees Witteveen