Renault creates a virtual duplicate of its factories to optimize them


Renault is picking up speed in terms of optimizing its vehicle manufacturing activities by creating a virtual duplicate of its facilities. A device that the manufacturer calls “industrial metaverse”. Renault uses Google to centralize its data and enable artificial intelligence algorithms to optimize its processes.

Global data centralization and artificial intelligence

A central platform brings together all data from factories, vehicle assembly lines and the supply chain. From there, the manufacturer applies monitoring and optimization algorithms based on artificial intelligence and machine learning.

All of our group-level devices work together digitally

Thanks to the unique structuring of our data, we are now able to scale and process data at the batch level » welcomes Renault. ” When others try to capture data, create and improve their digital twins [Jumeaux numériques], today, we were able to connect all of these, all of our digital twins. All our cluster-level devices work together digitally, this is called the industrial metaverse continues the automaker.

This is what Renault presented to the press on October 25 at its Flins plant. The presentation device brings together a dozen booths. This “mini show” also aims for the car manufacturer’s internal teams to communicate about its industrial achievements for two weeks. This ranges from data collection from the field, equipment and goes to algorithms for overseeing production chains and optimizing factory supply and customer delivery, as well as additive manufacturing technologies. and operator training using virtual reality.

Supervising a Renault assembly line in Douai (Presentation 25 October)

Significant financial benefits

The goals of transforming to a central virtual platform are lofty. This new approach should make it possible to achieve 320 million euros in savings on fixed and variable costs by 2025. Similarly, it should lead to the release of 260 million euros in savings by reducing inventories. ” Centralization reduces inventory because you know exactly what you have and where it is. It is not useful to have duplicates. This results in 20% to 30% savings Renault underlines.

Each object is modeled, identified, connected and has its own digital avatar

Also, a target is to reduce by 60% the time between ordering and delivery of a vehicle in front of the customer and contribute to reducing warranty costs by 50%. Renault Group has created a platform that unites with a single standard description all objects, screwdrivers, robots and even packaging. A standardized data model can be used to describe a pump whether it is in France or Spain. Each object is modeled, identified, connected and has its own digital avatar.

This constitutes the true-to-life digital double of all factories and the Supply Chain of the Renault group. There are thus 8,500 connected devices, 4,400 connected workstations and 10,000 energy sensors installed. Supply Chain tracks 90% of trucks in real time in Europe using Shippeo, and Renault tracks the delivery of vehicles to dealers, 700,000 of which have been tracked. 1 billion data sets are collected every day. “ This is very rich data, such as curves or photos Renault says.

Real-time simulation helps you adapt faster

The digital twin makes it possible to simulate in real time the entire ecosystem of Renault factories and the countless relationships between avatars to optimize their functioning in the physical world. Real time allows you to intervene in the right place at the right time. In the event of an unforeseen event, an accident or a disaster, real-time simulation makes it possible to adapt more quickly.

Computer vision startup Buawei can check if the right tires are fitted to vehicles

Artificial intelligence has been mobilized in many processes. For example, Renault works with Lille startup Buawei to identify on the assembly line whether vehicles are equipped with the same tire reference. This reference is written on the rubber, but it is difficult to see, hence the interest in using computer vision to perform the verification. In case the references differ between the tires, the vehicle is removed from the assembly line.

Renault has developed an image database with Google; called MP4 AI. This platform facilitates the development of new vertical solutions. Buawei’s solution, for example, is a vision control system that uses artificial intelligence models trained in Google Cloud and then executed in the field, the so-called “Edge”. Renault relies on Google’s devices, “devices”, to run algorithms locally in its workshops.

Checking Tire References Using Computer Vision (25 October Presentation)



Leave a Comment