I have been asked to contribute to a book on Digital Twins, and specifically to provide a Chapter on the Future of Digital Twins.
I have started to put together a few ideas and I would like to get feedback, as well as more ideas, on what the evolution of DTs in this decade will be like, so please, share your thoughts!
The concept and application of Digital Twins have been evolving rapidly in the last fifteen years and all signs point towards further rapid evolution in this decade.
The purpose of this chapter is to identify these signs and to follow the paths they are indicating to highlight potential application and issues.
In short the evolution has to be considered in terms of:
- increased capabilities
- broader fields of application
In the last decade we have seen, see fig. 1 an evolution of the concept of digital twin as a model of a physical entity (stage 1), mostly used in the design phase, to a mirroring of an existing physical entity (stage 2) that can be used as a reference model for a specific physical entity, thus introducing the aspect of “instance” (at stage 1 this was not existent, a digital twin would represent a turbine model, not a specific turbine that was manufactured). This has introduced in the Digital Twin the Thread component, i.e. the record of the evolution of that specific instance (how the physical entity was manufactured, how it has been customised, sold, …
A subsequent evolution, stage 2.5, led to a mirroring of the present status of the physical entity (here the concept of instance is crucial since several instances of a product will be in different states, depending on how they are being used). The means to synchronise a given physical entity with its associated instance could differ, from a mere data entry to a continuous flow of data from the physical entities, usually leveraging on sensors, IoT, embedded in the physical entity, connected through a network to the digital twin. This has added the Digital Shadow component to the Digital Twin (digital model, digital shadow, digital thread, completing the three component of today’s digital twins).
At stage 3, the one most of industry is in the adoption of Digital Twins, we have a bi-directional interaction between the physical entity and its Digital Twin. This resulted in an expanded Digital Thread, since now it is recording all significant events occurring in the operation (and maintenance) as well as in the capability of affecting the operation of the physical entity, potentially in an autonomous way.
This step signals a departure from the original concept of Digital Twin, where it was a passive representation of a physical entity. As soon as the digital twin can influence the physical entity it -potentially, can become an active component. Today, as represented in the figure, in most cases this does not happen. The digital twin “influences” the physical twin just by acting as a gateway for an external controller to operate on the physical twin. The interaction is not the result of an autonomous decision on the side of the Digital Twin.
The next step takes the digital twin to stage 4 where a certain degree of autonomy in the pair digital-physical twin is present. At the stage some of the functionality of an entity may be provided by the interworking of the digital/physical twin, local data analytics are used as well as a minimum of intelligence at the digital twin level. The level of autonomy increases as Digital Twins move at stage 5 where they can interact, as autonomous agents, in the cyberspace expanding the local data analytics to global data analytics.
On the horizon is the possibility for self-generation of digital twins, of hierarchies of digital twins and meshed networks of digital twins, all together giving rise to an emergent ambient intelligence (not represented in the figure).
In terms of broader fields of applications the evolution has been following an exponential curve: from the early application in manufacturing (leveraging on the availability of digital models created in the design phase -CAD: Computer Aided Design- to their adoption in construction, automotive, aerospace, cities and more recently in agriculture, healthcare, energy.
The expansion is progressing and it is now decisively moving towards the modelling of processes, of immaterial goods (economy, finance, knowledge), and people.
This latter is a natural extension of healthcare and of resource management starting to involve education, particularly new continuous education paradigms.
The expansion has been taking place horizontally, i.e. by involving more application areas, as well as vertically, i.e. by extending the coverage of the physical entities, leveraging on the increasing number and quality of sensors and the related increase in the quality and volume of data.
I am willing to bet that by the end of this decade digital twins will be a part of the landscape and used as the bridge connecting the cyberspace to the physical space becoming the pillars of the digital reality where we will be living and doing business.