Performance Assessment in Self-organising Mechatronic Systems: A First Step towards Understanding the Topology Influence in Complex Behaviours

Abstract

Abstract. The research and development of self-organising mechatronic systems has been a hot topic in the past 10 years which conducted to very promising results in the close past. The proof of concept attained in IDEAS project [1] that plug&produce can be achieved in these systems opens up new research horizons on the topics of system design, configuration and performance evaluation. These topics need to consider that the systems are no longer static prototypes but instead several distributed components that can be added and removed in runtime. The distribution of modules in the system and their inherent connections will then potentially affect the system's global behaviour. Hence it is vital to understand the impact on performance as the system endures changes that affect its topology. This article presents an exploratory test case that shows that as a system evolves (and the nature of the network of its components changes) the performance of the system is necessarily affected in a specific direction. This performance landscape is necessarily complex and very likely nonlinear. Simulation plays therefore an important role in the study of these systems as a mean to generate data that can be later on used to generate macro level knowledge that may act as a guideline to improve both design and configuration.

Automated experience-based learning for plug and produce assembly systems

“…Successful EU-funded research projects that developed multiagent systems and plug-and-produce technology include IDEAS (Onori et al 2012;Neves et al 2014b;Hasan, Onori, and Wikander 2014), GRACE (Pereira, Rodrigues, and Leitão 2013;Cristalli et al 2013) and PRIME (Antzoulatos et al 2014). In IDEAS, plug-and-produce self-reconfiguration is accomplished using only multiagent control programmed in intelligent embedded systems, without any higher-level coordination.…”