Multi-Agent Systems in Support of Digital Twins: A Survey

The joint use of technologies such as IoT, Artificial Intelligence, Cloud Computing or Virtualization has fostered the development of digital twins (DT). A DT is described as a physical entity, its virtual counterpart and the data connections between both. Digital twins are increasingly being used to enrich physical entities by exploiting different computational approaches, which are applied to the virtual twin part. One of such approaches is the multi-agent systems (MAS) paradigm. It is claimed they resemble DT in many features. In order to analyse the suitability of MAS for DT, this paper presents the results of a systematic literature review focused on the analysis of current proposals exploiting MAS to support the design of digital twins. We found that the integrating the multi-agent paradigm with digital twins can be challenging, because the distinction among them is sometimes blurry. Moreover, it has been detected that MAS are generally the interaction environment for the DTs, and data of the DTs allow agents' better decisions to be made in real time. That is, the massive volume of data stored by the DT allows agents to make decisions based on these data, and on the other hand, MAS shapes the environment where the DTs operate and interact.