Modeling resilient cyber-physical processes and their composition from digital twins via Markov Decision Processes

Cyber-physical processes are those processes in which (some) tasks are autonomously enacted by smart objects and have a physical effect. They are interesting in current Internet-of-Things (IoT) scenarios, in which the resilience of the overall process is crucial. Digital twins, widespread in smart manufacturing but also in many other novel scenarios, can be used as building blocks of cyber-physical processes. In this work, we focus on the orchestration of Digital Twins using an AI technique such as Markov Decision Processes (MDPs). We formalize stochastic composition of processes as LTLf goals, we present a proof-of-concept implementation and exemplify in an Industry 4.0 scenario.