The notion of “open control” encompasses classical explicit control and an innovative kind of control called “implicit”, developed to influence a set of entities instead of controlling each of them directly, enabling the control to be more agile and reactive. The considered entities are decisional, autonomous and cooperative ones (they are called “active/intelligent entities”). Multi-agent and holonic modeling approaches are then logically considered as possible modeling tools. For the interaction among the entities, several modeling approaches are used (direct messages exchange, contract-net, potential fields, reinforcement learning and stigmergy…). The focus is held on the open control of dynamic allocation of tasks and dynamic routing of entities processes (e.g., routing of products in a FMS or routing of vehicles in an urban traffic). Since the “active entities” must interact with their environment, emerging concepts, such as ambient intelligence will also be considered for HMI purposes or inter-mobile and heterogeneous entities communications.

In the context of production where products are in-progress and not fully operational, we are working on the concept of “augmentation”, that is an associated set of functions for products and relevant technical solutions (embedded/distant) that enable them to be considered as active, even if they are under construction.
 

Applications concern:

• Transportation systems (railway vehicles, SURFER project or urban traffic, PPF “Coeur de ville” project),
• Healthcare systems,
• Manufacturing and “intelligent manufacturing systems” (AIP-PRIMECA flexible manufacturing system).
   

Potential field-based simulation of the open control concept using NetLogo.

Heterarchical control of the AIP-PRIMECA cell using the open control concept (intelligent shuttle fleet)

Self-organized Road traffic simulator and associated KPIs

Robust control of complex systems

This second theme focuses more precisely on the high ill-structuration of data that may occur in some contexts making it hard to use usual deterministic sizing, scheduling and performance analysis approaches. For example, in the case of curative maintenance, the estimation of the duration of the task cannot be predetermined and can strongly vary (e.g., in high speed TGV trains). It is also the same in healthcare systems when surgeons operate. More, in such systems, emergency tasks occur regularly. In such an ill-structured context, the proposed research activity is mainly led on sizing and scheduling processes. The approach consists in building robust sizing and scheduling, mainly based on meta-heuristics (in particular, genetic algorithms and robust evaluation function) and discrete event modeling and simulation (for stochastic modeling and verification purposes).

Applications concern mainly the healthcare systems:

• An integrated decision support system (with University of Westminster and Valenciennes public hospital),
• Framework for the assessment of operating theatre (with Tivoli and Bruxelles Belgian public Hospitals),
• In-Hospital resuscitation problems: localization of monitors and defibrillators, organization and optimization of emergency medical team interventions (with Maubeuge Public Hospital).