Fifteen years ago at the Future Centre we worked on the idea of a “digital shadow”. We though that as bits were more and more able to “mirror” atoms we would see a growing creation of bit images, virtual images if you want, in the Cyberspace.
Having a virtual image would become useful in several situations, like having a Wayback time machine of yourself. Did you forget where you left the keys? Rewind and see yourself leaving them on the chest of drawers. Did you break a leg? Show your virtual self walking to the doctor taking care of your rehabilitation so that it can be customised to you… We came up with many examples of potential applications of a digital shadow. We discussed the idea with the top management and … we were stopped: too many privacy implications, it was not worth it.
Now it has happened, and it is going to happen even more in the coming decade. The name changed, it is not a “digital shadow” but a “digital twin”, a better name, I have to admit, but it is basically the very same idea.
It has started in areas where there is not privacy concern, like the mirroring of engines (see the clip).
General Electric is probably the first company that came up with the Digital Twin name and most importantly with an industrial application of the concept.
They are using digital twins to improve the efficiency of their power turbines (generating electricity from steam and from wind) and more recently the efficiency of airplane engines.
It is not an experiment, with some prototypes. As of the end of 2017 they have some 551,000 digital twins that correspond to products, parts of products, processes and systems. They are stating that Digital Twins are changing the industry, even more: they are seeing digital twins as a step in the direction of a symbioses between human minds and machines.
Tesla is applying, basically, the same concept to its cars. Every car reports back on a daily bases its day experience and those data are used by simulation programs on the digital twin to discover possible anomalies and provide corrective actions. The multitude of digital twins, for Tesla as for GE, makes possible to learn from multiple experience. Tesla declares that it gets the equivalent of 1.6 million miles “a day” of driving experience and this is fed back to each car in a continuous learning process.
A city is a very good candidate to have a digital twin. Here imagine to have its various systems (transportation, power lines, telecommunications, health care, logistics…) modelled via digital twins. You would be able to simulate much better on the city’s digital twin what may happen and take better decisions. And mind you: the ever growing presence of IoT (see my yesterday post “Towards a sensors infrastructure“) is actually providing the needed connection between a city and its digital twin. This is a topic that we will be discussing at the coming WF IoT 2018 conference in Singapore on February 6th.
There are issues that go beyond technical aspects in this interplay between the physical entity and its digital twin. Think about the digital twin of an airplane engine. It can detect an anomaly and therefore it will instruct the Flight Management System to decrease the power of that engine to avoid heating and wearing and to increase the other engine power output plus a reconfiguration (trimming) of the wings to balance the “unbalanced” trust now provided by the engines. What about the pilot? All of this is going on under his seat with very little awareness on his part… Is he still responsible of the plane or the digital twin has taken over?
Examples given so far are interesting but are not about a potential “our” digital twin. Yet, I was discussing with some researchers at Nokia Finland about Digital Twins and they told me that the growth of wearable sensors, smart watches to name but one example, is providing a growing set of data that can establish a connection between a person and her digital twin. And if you care to make a few steps into the future, just into the next decade, you’ll see that more and more people will have their genome sequenced, and, as I will discuss in tomorrow post, there is a growing possibility to analyse their genome to make predictions that can be tuned using the data coming from a smart watch. This is clearly raising much more thorny issues, in terms of privacy and external control. It is indeed casting a shadow. Perhaps our naming of digital shadow was not too far off…