Enabling human–CPS cognitive interoperability: Cognitive architectures as technologies for human-like cognitive digital twins

Cognition, the set of mental processes that enable humans to perceive, reason, learn and decide, plays an essential role in effective collaboration between humans and Cyber-Physical Systems (CPSs). To achieve seamless cognitive interoperability between humans and CPSs, it is necessary to integrate a Cognitive Digital Twin (CDT) and a Human Digital Twin (HDT) to provide digital representations of both physical assets and human cognitive states. In this article, we first analyse the three essential functions of CDT and HDT: emulation, cognition and simulation, and review the state-of-the-art technologies for each of them, from supervised learning and knowledge graphs to deep reinforcement learning. Focusing on the cognitive layer, we review the state of the art in cognitive architectures, describing their symbolic, sub-symbolic and hybrid types and reporting on their real-world implementations in different domains. We then assess the relevance of these architectures for the integration of human-like reasoning in CDTs. Finally, we identify the main technological challenges and gaps that need to be addressed in order to implement fully operational CDTs. • Introduction to the concept of cognitive interoperability in Human-Cyber Physical System collaboration and the use of Cognitive Digital Twins (CDTs) as a reference framework. • Decomposition of CDTs into three functions and review of state-of-the-art technologies. • Study of cognitive architectures, theoretical foundations and real-world applications. • A roadmap of challenges and gaps to be addressed for the industrial deployment of CDTs.