Research Projects

Digital Twins (DT) are emerging as one of the most crucial amalgamations of modern technologies, defined as a virtual representation of a physical system (and its associated environment and processes) that is updated through the exchange of information between the physical and virtual systems. Digital twins, being virtual, create the means of improving the performance of their physical counterparts by leveraging their computational power, data availability, and artificial intelligence, unlocking the potential to optimize control, predict future behaviour, support decision-making for optimal use and so much more, under one umbrella. As a comparatively younger concept, major efforts around the world are made to develop complex digital models for specific applications. However, challenges for the large-scale adoption of DTs are yet to become mainstream. Some of these scientific challenges are limited training data, time and effort; unauthorized access and manipulation; data privacy protection; data security; and trustworthiness. At a fundamental level, the project ambitiously aims to address these scientific challenges and develop a systematic methodology that pushes the fundamental possibilities of the digital twin as an entity by exploring options to research questions like: “How can a digital twin be trained with minimal training effort?”, “How can digital twins collaborate to arrive at common decisions within and across organizations preserving data privacy?”, and so on. Through the development of two functional laboratory setups mimicking manufacturing operations, together with developing mechanisms of collecting relevant data, unsupervised identification and modelling of events are aimed to be achieved. This would form the base for autodidactic training of digital twins. Extracting and storing these insights as knowledge, along with the integration of goals and constraints will endow intelligence to the twins making them capable of descriptive, predictive, and prescriptive analytics. Expanding the framework to comply with data privacy and security protocols, the twins can participate in collaborative decision-making with other twins without sharing data with trustworthiness and accountability. Finally, by implementing the methods as an end-to-end pipeline in manufacturing and healthcare scenarios, the versatility of ADVAIT digital twins will be validated. Thus, developing self-learning (autodidactic), versatile, accountable, intelligent, and trustworthy digital twins for collaborative decision-support Servitization across large-scale applications is the primary goal of the project.