Collaborative routing of products using a self-organizing mechatronic agent framework—A simulation study

• An innovative agent-based architecture for dynamic routing is detailed.
• The architecture is tested under simulation and the results are discussed.
• Results suggest that the fully distributed approach provides a feasible solution.

Scheduling is a fundamental activity in modern shop floors. It is also known to be a highly complex problem which has motivated several sub-formulations and the subsequent models. Traditional approaches, typically enumerative or heuristic, struggle to contain the computation complexity and often present solutions for restricted cases that feature unrealistic assumptions in respect to the system size, flow of products and the system logistics/behaviour. The multiagent-based architecture presented in this paper is aligned with a set of emerging architectures that seek to explore more heterarchical decision and control models to circumvent the limitations of the traditional approaches. The main distinguishing factor of the proposed architecture is that it directly addresses (re)routing/local scheduling of products in plug and produce systems. It does not make any assumptions on the alignment of the orders and, instead, it dynamically handles the potential rescheduling of the orders already on the system based on the available resources, and their state, in a time efficient way. The architecture was tested under a simulation environment, that is geometrically accurate and that supports plug and produce in runtime, to characterize its performance under dynamic conditions.