The Covid-19 emergency has made us realize that new technologies will play a key role in building the post-pandemic world. And as we consider which intelligences will help us realize a safe, sustainable and forward-looking path, the return of "relationships in presence" has already taught us that working together allows us to look to the future more optimistically. Among the most advanced sectors of Italian industry, automotive, despite the impact of the pandemic, has not slowed down its technological development processes. Connectivity, automation, electronics and shared mobility are the elements that will characterize aspects and scenarios of this industry in which the Digital Renaissance is exploring its full potential.

Marko Bertogna at the head of HiPeRT Lab, which now has some 70 researchers and doctoral students and is working on two major issues, namely the development of next-generation microchips and the scope of application of processors in new automated driving systems, a current and interesting field in which to make a revolution that goes hand in hand with humans and is driven by the talents of individuals, knows this well. But it was not always successful: at the beginning of this experience, companies were not particularly attracted to the lab's proposals.

"It was then that we began to question what processors would attract the attention of businesses in the near future. It all started with Moore's Law: a rule of thumb that allowed us to gradually reduce the size of transistors and allowed us to increase the number of transistors to be integrated within the same chip. The architectures in our lab, for example, have a number of transistors comparable to the neurons in a mammalian or human brain. The problem is that these technologies are increasingly expensive: the cost of the machinery that can produce them amounts to several billion dollars, and since only a few companies can afford to manufacture these kinds of chips, the number of players in the global market has shrunk significantly.

These transistors are multi-core, that is, they have the ability to have a large number of processors within the chip. In fact, today we are in the era of parallel architectures, a feature that allows for computational performance that until a few years ago was available only to certain computing centers.

In the automotive field, there are a hundred processors scattered throughout the vehicle. A profoundly complex system that requires computer experts, electronic engineers capable of making cars something increasingly far removed from the mechanical. Not surprisingly, cars are among the devices that will undergo the most change in the near future. The whole electronic part will be revolutionized and will go from hundreds of cores scattered around the vehicle to two, to three main controller domains. A process that will happen in all industries, from robotics, to industrial automation, to avionics.

As fewer and fewer providers are able to produce these platforms and it is very expensive to make them, it means that new chip designs will not be application-specific, but it will be increasingly common for the chips used in our mobiles or smartphones to also be configured for the autonomous vehicle, drone or robotic arm. In realizing this kind of software ecosystem, it is extremely important that it offers industries guarantees of security and predictability, which are critical in a system designed to move in the public realm, where we interact with users, such as an autonomous car or a home delivery robot. Already today there are intralogistics companies that have extremely advanced automated warehouses and more and more will continue to implement these systems, these new architectures that we need to learn to use to develop somewhat more futuristic applications.

What we do in our lab is try to merge two worlds: a world that needs to eat a lot of data, what is called data crunching, that is, processing data from a variety of sensors. Our vehicles are equipped with various cameras, laser scanners, etc. so we do what is called sensor fusion on board, which is the merging of data and signals that come from different sensors and require a lot of computation. On the other hand, we have the need to develop the applications that we are going to place on these vehicles and that cannot be the same as Apps developed for a game, for mobile or the web. These are safety-critical devices that need granular controls, very low latency erroneousness, for example, predictability-what-most-happen-in-the-worst-case issues of qualification and certification. All this is too much neglected in IT courses, where there 's a big unpreparedness, but I think we Italians could say a lot because even abroad they are not that advanced on these issues. We have before us the opportunity to seize this revolution, at the system level, at the application or inventive level, imagining what can be done.

It stands to reason that anything that moves, on a wheel, on a wing, on a propeller, on water, can be autonomous, and if it can be made autonomous, all indications are that it is likely to become so. If I don't need a driver or an operator and the same tasks can be accomplished automatically, it's better to gear up early, develop the technologies, hire the resources that will lead to the company being more ready for this industrial revolution.

Our industrial ecosystem has been very successful at the mechanical level, but it is somewhat unprepared to ride this technology-information revolution. The challenge is to intercept these trends and figure out how to translate them into our daily production realities. Increasingly, autonomous systems will have recognition capabilities, to replace those often repetitive and uninspiring tasks that are performed by humans today. This new computational paradigm, these chips capable of performing tasks that until recently were unthinkable to solve automatically are a small preview of how the world of the future may work. Anticipating trends about what will be autonomous and preparing in all domains that may be touched by this revolution will be the challenge to be overcome."

Download pdf article