The capability of a production environment to dynamically cope with changes of the production requirements is seen as one of the key enabling factors for highly automated and competitive production systems. In this regard, Reconfigurable Manufacturing Systems (RMSs) are endowed with a set of reconfigurability capabilities that can be related either to the single component of the system or to the entire production cell and the system layout.
In this context we get involved in the GECKO project (GECKO – Generic Evolutionary Control Knowledge-based mOdule) for developing a distribute software architecture composed by autonomous agents. Agents exchange information about the environment and cooperate in order to define and support the production flow of the shop floor. Information exchanged allow agents to build an abstract view of the shop floor they are deployed to. Each agent stores these information into its own Knowledge Base which provide an high-level and “functional” (by means of Ontology of Functions) view of the production environment. Then we have applied Flexible Timeline-based Planning & Scheduling technology to make agents able to dynamically adapt their capabilities according the their internal state and the configuration of the production environment [ICAPS-PlanRob 2014]. In particular this application scenario gave us the opportunity to start studying the relationship between the Knowledge Base representation of the agents with the Timeline-based Planning model we use for plan module’s activities. An important challenge we are currently pursuing is to define a high-level control loop which allows to dynamically infer the Timeline-based control model directly from the Knowledge Base.