To conclude, some remarks on MAS applications to manufacturing planning, scheduling, and control problems can be made. Throughout the chapter, two main features emerged for MAS approaches:

• MASs represent a means to implement decentralization in MS, i.e., to decompose the decision problems relevant to manufacturing production into a set of less complex subproblems tackled by one or more decision entities.

• MASs entail a flexibility that provides the preceding decision entities with autonomy and proactiveness, if needed. Regardless of the design choices adopted for an MAS to adapt it to a specific MS application, the MAS approach seems the best current technological solution to introduce agility in manufacturing.

In particular, recall the MS agility requirements, from the bottom up:

• Agility requires a production control model, which is distributed and highly reconfigurable, in order to respond to the modern requirements of (1) decentralizing the control procedures, currently embedded in PLC, directly on manufacturing workstations; and (2i) introducing the current facilities of high-level software (such as the reusability in object-oriented programming) into control software components. Agent or holonic systems satisfy both requirements because they clearly enable the decentralization of the control of single shop floor devices through the assigning of each to a single agent or holon that, in turn, is implemented according to recent software engineering specifications and allows simple software component reconfigurability and reuse.

• Agile scheduling demands the ability to alter decisions dynamically about production activities in order to react rapidly to changes in the supply chain; on-line scheduling and...