IEEE Future Directions
Related Articles
The automotive sector has adopted Digital Twins technology in the manufacturing to mirror robots in the assembly line. In the last few years it has started to create and use digital twins of the product, vehicles produced. More and more automotive companies are nowadays equipping cars (and trucks) with IoT and receive a stream of data reporting the various car components status. The stream of data often includes location data and this may create an issue of privacy. Some manufacturers, to avoid this type of issues keeps the data record in the car and these data are only harvested when the owner requires a car check up (some of the data ends up in the key fob). In this case all data analytics take place in the car.
A whole new class of IoT and AI supporting chips designed to provide intelligence at the edge (like the STM 32 series) is now enabling local intelligence and support for local operation of a Digital Twin. This local intelligence would be able to both signal an emergent issue to the driver as well as to report the problem to a service centre for proactive maintenance, possibly on the fly without disrupting the service (the DT may take action, autonomously or guided by the service centre -a software application-, to alter the vehicle parameters thus ensuring that it can keep going deferring the required maintenance to a later time).
For the time being, however, and to my knowledge, only Tesla has a real digital twin associated to each one of its vehicles. Daimler (trucks), Porsche and Mercedes (DT used in production and for Formula 1 cars) are hinting at adopting DTs for their products.
Tesla has embraced Generative Design, an evolution of CAD -Computer Aided Design- that uses AI to optimise the design studied by the engineers based on the goal. Associating AI to the Digital Twins of the cars already in use (to the data provided by those DTs) it is possible to take into account feedback from the “operation” field, in the true spirit of Industry 4.0. Each Tesla car is associated to a DT and that DT is reporting back to Tesla GB of data every day. This avalanche of data is analysed through AI (and Machine Learning) resulting both in monitoring, offering of services, and fine thing of production. Since Tesla cars are basically computers with specialised software it is possible to update the software whenever needed, both to fix glitches and to offer new features. Here again we are seeing the convergence of product, service and knowledge.
Data collected by Tesla from their cars are massive, an estimated equivalent of 3 billion miles of data are now on their servers enabling unique (in the automotive market) data analytics, AI and ML. Consider that a car may generate a few TB of data each single day! No surprise that some analysts look at Tesla not as an automotive biz, rather as a data company.
By far Tesla is using DTs at stage 3, however there are a few nuances that show a use that would place them at stage 4 and even 5.
