
As artificial intelligence is progressing we can shift from pure syntactical interactions to semantic based interaction. This is what happens with the newest robots that are more and more aware of:
- their surrounding
- the problem space they are addressing
- the goal they pursue
- the” help” they can get from other resources (first of all other robots)
This evolution towards “cognitive” machines started in the last decade and is now accelerating.
Let’s take a back-step: in the last decade more and more industries turned to Digital Twin as a way to improve their production. By having a digital mirroring of their workshop resources they could simulate different ways of using them, monitor them and eventually guide them from the cyberspace. These digital twins were (are) bridges between the physical entities and their mirror images in the cyberspace. Some functions could actually be performed in the cyberspace with no need to involve the physical entities thus increasing the overall efficiency and providing more flexibility. Over time (and this is still ongoing) we have seen that some AI functions that made use of Digital Twins (data analytics, simulation…) are now becoming embedded in the Digital Twin itself. This, in way, is changing the definition (and concept) of Digital Twin since now it is no longer a digital replica of a physical entity but has some functionality (smartness) that is not part of the physical entity (more of this in following posts).
This situation brings the Digital Twin into “stage 4” where we no longer have a separation between the physical and the digital twin: they are both essential to define the entity that would no longer be able to perform (at the same level) if only one of the two were present. In order for the Digital Twin to become a component of the entity it has to be aware of the knowledge space of the physical entity (that includes its awareness of the operation space, of its environment). This mirroring of the knowledge of the physical entity and the expansion of the knowledge characterises a Cognitive Digital Twin. As mentioned the Cognitive Digital Twins were “born” in industrial environment, with possibly IBM spearheading this evolution (2018).
Notice that a Cognitive Digital Twin (in the context of an Industrial application – I’ll turn to human Cognitive Digital Twin in subsequent posts) is more than just a mirroring of knowledge, it is an extension of knowledge: in other words an entity having a Cognitive Digital Twin has a knowledge space that is in part within the physical entity -mirrored in the Digital Twin- and in part extended by the Cognitive Digital Twin. This latter can have access to tons of knowledge available in the Digital Space and -most crucial- can derive applicable knowledge to the context of interest for the physical entity. This derivation of knowledge to make it relevant and applicable requires intelligence, that is why Cognitive Digital Twins require AI.