Siemens accelerates towards the autonomous factory, with a providential ride from artificial intelligence
A robotic arm will always be a robotic arm, but at Siemens it is a robotic arm that serves as a business card for something bigger that goes by the name of an autonomous factory. The robotic arm that many people think is nothing new was used by Norbert Gaus, vice president responsible for Research and Development at Siemens, to show the potential of artificial intelligence (AI) when it comes to carrying an automaton picking up different objects. For a human it’s easy – for Siemens it’s proof that it’s feasible to create machines that discover the best way to perform a task. The technology debuts on the Xcelerator platform this summer. For Siemens, it is the first step towards autonomous factories. And that was the message conveyed at the Siemens AI with Purposes event in Munich.
âEste The robotic arm was trained to discover the optimal points that allow it to grasp each of the objects,” describes Norbert Gaus.
At Siemens, industrial controllers are a historic business that is carried out under the Simatic brand. These controllers work as modules that determine how different equipment and robots should behave in different scenarios and functions.
Today, a third of the controllers used in industry around the world are produced by the German brand, but the technological leap presented by Gaus during the Munich event promises to change a large part of the routines in a factory with robots. Until now, whenever a robot needed to deal with a new object, it was necessary to insert programming codes to change the different functions. But with Microsoft’s generative AI resource, it became possible to create a “copilot†that receives new tasks and functions with instructions in the mother tongue, and even more importantly, allows the necessary data to be extracted in advance. from cameras to learn how to perform new functions, such as grasping objects of different shapes and weights.
This is why the robotic arm that Norbert Gaus showed does not need to be programmed for each object it handles. “These robots don’t have to be suddenly trained to handle new objects, and they can pick up anything.â€
These are small movements of a robot that promise to transform entire factories around the world, Siemens predicts. “We are taking the first steps in what we call adaptive factories… For example, we can decouple a functionality from the hardware. In the past, any more specific functionality was linked to a specific hardware component. And this is what we are decoupling,” describes Norbert Gaus.
All AI has an impact on the job market… and it was through the mouth of one of the partners that Siemens brought to the event to publicize the latest technological news, that one of the most enlightening predictions on the topic emerged: “AI will not will replace humans, but humans who use AI will replace humans who don’t use AI,” said Matthias Kuss, director of Data Solutions at Fresenius.
It was also from this executive that the most practical of industrial parables about AI emerged: “This technology is like toothpaste: once it comes out of the tube, it doesn’t go back in. .†Equally metaphorical, Norbert Gaus summarizes evolution in one sentence: “We may not be able to see it, but AI is everywhere.â€
If it was already true that almost no factory can afford to send robots away, then it will be even more true that no factory will be interested in ending the use of robots that have their own intelligence to perform different routine functions.
Instead of facing negative evolution, Norbert Gaus envisages an increase in the level of functions that will be assumed by humans. “The ‘copilots’ will free us from a lot of routine work that no one really wants to do, and spend more time on the more complex problems,†he says.
In the field of complexity reserved for humans, cases involving moral issues, emotions, unresolved dilemmas, or options that follow the rules, but suddenly turn out to be wrong due to changes, can easily be seen. that of contexts, which humans can easily uncover, but machines, which have no experience of their own, can hardly evaluate.
Sonja Zillner, leader of the AI Trust area, gives as an example a simple certification about something that machines already do: to date no certification has been created that a machine can really classify and distinguish a dog from a human, or a car from a bicycle, among many living beings and objects that surround us on a daily basis.
This certification will certainly be crucial to guarantee that a project like the one in which Siemens has participated in the development of a train without a human driver. Sonja Zillner confirms that machines and automatons will have to know how to apply the new EU AI Regulation rules, but also recalls that there are stratagems that have been applied to ensure that greater capacity to deal with with the variability of everyday life.
“I have colleagues who train AI systems with ‘dirty’ data (which has errors or is misleading), so that they can prepare them for scenario changes,†says Sonja Zillner.
Norbert Gaus, vice president of Siemens, considers Artificial Intelligence one of the pillars of autonomous factories
Primary Eskipehlivan
More than once, Siemens representatives reminded us that AI is not exactly new. On one occasion, there were even people who said that the German brand registered between 40 and 50 patents related to this technological area.
Still, there is the notion that, these days, the power of the algorithm will be somewhere between the big American tech companies and the new startups that have been launching so-called AI models that extract information. text, images and sounds. And for this reason, the strategy always involves partnerships with Microsoft, AWS (Amazon), and also other smaller partners or open source solutions, which allow you to understand how it works and carry out adaptations.
Michael May, responsible for AI at Siemens, predicts, in the near future, the emergence of smaller generative AI models that can be processed within a factory – or at the limit by a single piece of equipment – €“ without having to use the Internet to be trained or gather more information.
Matthias Kuss, executive at Fresenius, recalled that the Artificial Intelligence movement is unstoppable
Siemens
The range of bets includes simulation tools that allow us to know how an object behaves in the face of different adversities, but also platforms that help design products depending on needs, requirements or usage scenarios. In other cases, it is the AI itself that gains self-analysis capacity and presents potential solutions for the flaws and errors it discovers.
“We will be able to have better AI responses in the future, but the objective is to always keep the human involved: even for legal reasons,†emphasizes Michael May.
AI is the acronym of the moment, but Gaus emphasizes that it is just a piece added to the puzzle with which he intends to pave the way for the autonomous factory. “It is also necessary to ensure that there is full connectivity and interoperability (between different families of equipment). Cybersecurity must be guaranteed permanently. It requires very strong synchronization and digital twins (which replicate real objects). Therefore, it is necessary to combine many technologies to reach an autonomous factory,” responds the vice-president of Siemens.
Today, humans and robots already share work spaces. As in the parable of the toothpaste, no setback is foreseen, but Norbert Gaus also rules out any need to create more rules to harmonize these two groups of workers.
âThe regulations are already very strict. And it is very clear when it comes to bringing humans and machines together in the same space to work, while respecting current regulations. But we don’t need more regulation,” argues Norbert Gaus. Technology will have to manufacture the future.
Expresso traveled at the invitation of Siemens
