The roadmap of digital twin evolution defines V stages (more might be added as evolution progresses):
- the Digital Twin is used to model the physical entity, as example during the design phase or to document a Physical Entity
- the Digital Twin mirrors the Physical Entity and is used, during the physical entity life cycle, for simulation/emulation
- the Digital Twin and the Physical Entity, also called the Physical Twin, are in synch because the Digital Twins receives status information from its Physical Twin. This information is often in form of data created by embedded IoT. The DT may also receive the flow of I/O signals exchanged by the Physical Twin with its environment.
- the Digital Twins is responsible for some of the functionality delivered by the Physical Twin. At this stage there is an overlapping between the two and it is no longer possible to keep them apart since the availability of the DT is essential in the behaviour of the Physical Entity in order to deliver all functionalities.
- the Digital Twin is autonomous and becomes a superset of the Physical Entity. It still mirrors the Physical Entity but has additional information/capabilities.
When we use this roadmap for Cognitive Digital Twins it changes as follows:
- there is no Cognitive Digital Twin at this stage, unless one would consider as such the target knowledge to be achieved by an education curricula
- the Cognitive Digital Twin is used to represent the acquired/actual knowledge space of a person/organisation (like a cv of a person)
- the Cognitive Digital Twin is used to represent the acquired/actual knowledge space of a person/organisation (like a cv of a person) and there are means to keep this image “up-to-date” by tracking the evolution of knowledge. In this case, the CDT can be also seen as a prosthetic (I forgot something and my CDT can step in to provide the knowledge I forgot)
- the Cognitive Digital Twin, in addition to the knowledge space of its Physical Twin, owns additional knowledge space, acquired through a variety of means that are part of the creation process of the CDT (like access to a specific knowledge space, access to knowledge services…). This additional knowledge space augment the knowledge space of the Physical Twin that can use it in a seamless way
- the Cognitive Digital Twin has the capability to autonomously expand its knowledge space and makes it available to its Physical Twin. Additionally it can share its knowledge space, according to a defined framework, with other CDTs.
Obviously, the usefulness of the CDT increases with its evolution. Let’s go into the details.
At stage 2 the CDT contains the representation of the knowledge of its physical entity, based on an ontology that describes the meaning of each single knowledge “entity”, like OFDM coding. KaaS, the tools developed by the Digital Reality Initiative contains over 11,000 such entities. Each entity can be connected to other (sub) entities (like 5G is connected to “modulation”, “base station”, “network slicing”…) and can be connected to semantically close entities (like 5G can be connected to 4G…). A knowledge entity may describe a skill, an experience… Additionally, the CDT can (should) include the thread representing the evolution of the knowledge / experience. As an example the 5G can be represented in a thread that mirrors the evolution of knowledge of the physical entity (the person, the organisation) over time. A thread is also used to represent the sequences of experiences gained by being involved in activities, in companies…
The CDT needs to be periodically updated to reflect the evolving space of knowledge of that person/organisation.
One can use this CDT to get a glimpse onto the knowledge space of a person/a team/an organisation. It can also be used by data analytics and AI application to determine the knowledge gap between that CDT and the desired knowledge space (this can also be described using a CDT, and it would be a case of CDT at stage 1).
An organisation can create a CDT that is a cluster of CDTs in that organisation, like a project team CDT can be represented by the cluster of CDTs of its component, plus the knowledge space provided by the machine/tools used by the project, plus the knowledge connected via external consultant, supply and delivery chain.
At IEEE work is ongoing to create CDTs for each IEEE Society. These CDTs can answer questions like:
- should I join this Society if I need to access this knowledge space? Or, what IEEE Societies contribute to this knowledge space?
- what overlap exists among any two given Societies?
- which Societies should be involved in a given initiative? Or what knowledge asset can a given Society bring to this initiative?
These questions can be applied to CDTs of any physical entities, such as advanced robots and AI software. In the coming years the quest for transparency in AI will bring these aspects to the front stage: what is this application knowledge? How is it applied to the problem at hand?
… more to follow