Application of multi-agent planning to the assignment problem

Nowadays, a globalization of national markets requires developing the flexible and demand-driven production systems with new innovative concepts of management, information processing, production scheduling and planning. The presented work focuses on the low-level planning, where the multi-agent solution towards a job-machine assignment is considered. The main point of the discussion is the flexibility of planning systems ensured by the concept of agent's roles and emergencies. Depending on the state of emergency, the system receives stepwisely additional degrees of freedom to adapt the planning to the changing conditions of the manufacturing floor. The distributed constraint satisfaction and optimization approaches, underlying the suggested method, as well as activities of rescue agents, are described in the form of Petri networks providing both the conceptual notions and main details of implementation. (C) 2004 Elsevier B.V. All rights reserved.