By akademiotoelektronik, 29/12/2022

AI, robotics, cobotics, predictive maintenance

The last number

Regularly placed in the headlines, is AI just a fad or does it herald a paradigm shift that will upset our daily lives beyond our factories? The answers to these questions are often different according to the interlocutors. We still need to agree on the definition of AI. According to Jacques Baudron, the founder of Ixtel and teacher in academia and engineering school: "The weak AI available today is that of an automaton that can only make decisions by drawing from what is available to it. has previously been instilled; the strong artificial intelligence that we hope for tomorrow is that of a human being who decides with his conscience and his emotion”.

Same observation from Jérôme Laplace, founder and director of the Girondin companies Génération Robots and HumaRobotics, who chairs the Aquitaine Robotics cluster: “We are still at the beginning of AI. AI today is based on Machine Learning and Deep Learning, which form the basis of the new paradigm of computer programming. We are still a long way from a robot that develops a consciousness of its own.”

An artificial wave

Even if we are only at a weak application of AI, we are witnessing a real surge that will radically change the way we produce and profoundly modify our behavior. The numbers speak for themselves. Between 2015 and 2018 alone, the introduction of AI in companies increased by 270%: this is what emerges from the latest study by Gartner following a survey of 3,000 industrialists present in 89 countries. . “We have never invested so much in AI technologies as in recent years.”

According to Gartner, this increase is partly explained by the fact that AI technologies have reached maturity. “If a company is not using AI these days, there is a good chance that its competitors will do so and gain an undeniable advantage,” said Chris Howard, vice president at Gartner. On the side of filing patents, it is the same enthusiasm: the report published by the World Intellectual Property Organization (WIPO) on technological trends 2019, devoted to AI, reveals that the number of "patents in intelligence growth of 28% on average per year between 2012 and 2017”.

A total of 55,660 patents were filed worldwide in 2017. 89% of patents relate to Machine Learning with exponential growth in Deep Learning, which works with clusters of artificial neurons. “This growth is fueled by the wealth of existing digitized data and ever-changing computing power, with a potentially game-changing effect: by detecting targets among billions of seemingly unrelated data, AI can improve predictions and enable increase productivity and performance," the report says.

A quantum leap since 2013

In addition, Deep Learning, which has enabled the development of visual recognition, stands out from other Machine Learning techniques by the very short time between the moment when scientific studies are published and when patents are filed. . This period is generally 10 years for other Machine Learning techniques: for Deep Learning, there is almost no longer any lag. Francis Gurry, Director General of WIPO says about AI: "We have seen a quantum leap since approximately 2013. Our impression is that we are not going to see a decrease in public interest shortly. The phenomenon is so deep, and even if many applications will be invisible, I think we will still be talking about social effects for a long time. »

In this frantic race to register patents, 5 major groups occupy the upper hand. The first is IBM, which tops the rankings with more than 8,000 patents for all functions. The American giant has just announced that it will devote 2 billion dollars to the creation of a new research center on the campus of the Suny Polytechnic Institute in Albany, New York. This new hub will focus on research, development, prototyping, testing and simulation of AI-based computer chips. “Once established, the AI ​​Hardware Center will become a global hub for innovative research and development,” he announced earlier this year. Behind Big Blue, Microsoft is in second place with 5,930 patent filings followed by Toshiba, Samsung and NEC, respectively holders of 5,223, 5,102 and 4,406 patents. Toshiba and Samsung are the 2 firms that have filed the largest number of patents in the field of visual recognition.

Mature technologies

This explosion of investments and patent filings is linked to the coming of age of AI technologies. This is particularly the case of visual recognition in full deployment within companies. The first successful system called AlexNet was developed by Geoffrey Hinton (nicknamed the godfather of AI) and his team at the University of Toronto in 2012. From this period, startups and industrialists have developed applications increasingly efficient. This is particularly the case of Vinci Energies, via its Actemium branch, which has developed a system to inspect the fuselage of aircraft in Toulouse. Developed by Diota, a French start-up (specialist in augmented reality) which designed its technologies with the Commissariat for Atomic Energy and Alternative Energies (CEA), the solution is based on the use of a AGV (automatic guided vehicle). The autonomous vehicle takes photos of the aircraft which are then processed to measure the differences between the model and reality. British astrophysicist Stephen Hawking, who died in March 2018, has repeatedly warned of the risk for humanity of seeing machines become much smarter than humans: visual recognition is now one of the first areas in which the machine is able to replace the human.

Optimize decision-making

To face an increasingly complex globalized market in constant change and to stay in the race, companies must innovate, produce and supply more, by optimizing their flows and limiting their stocks in order to gain in efficiency and in agility. They must deal in real time with the various fluctuations that affect them. In this context, Machine Learning and Deep Learning are becoming valuable allies that will allow them to simulate, predict and make the right decisions. They have become essential for ERP and other MES publishers, the software packages that integrate them becoming real performance accelerators. In addition, the development of "communicating" ERPs allows companies to have the means to improve their economic intelligence capacity, to supervise their production line in real time, to carry out preventive maintenance & predictive from anywhere. With these new analysis tools, manufacturers have a global view of their entire supply chain, which helps them make the right decisions and adopt the right strategy at the right time.

Gains in agility and resilience

“For manufacturing operations, AI-enhanced predictive maintenance enables better anticipation of equipment failures by combining data from Internet-of-Things (IoT) sensors and other machine indicators. This can increase asset productivity by 20% and reduce overall maintenance costs by 10%,” says McKinsey.

For the publisher VIF, the use of AI in anticipation software allows manufacturers to make the right decisions at the right time and to anticipate failures as well as to establish strategic plans on periods covering 18 months. In order to be able to offer more effective solutions, the Toulouse company AGILEA created in 2017 the joint AGIRE laboratory (contraction of agility and resilience of companies) with IMT Mines Albi and Armines.

"In this laboratory, we use all Deep Learning technologies to develop tools for collecting, analyzing and processing data so as to have simulation models that will allow teams to choose the best solutions and best alternatives to gain agility and resilience,” argues Philipe Bornert, its CEO.

For its part, Supratec, a specialist in the design and sale of innovative industrial equipment, has just opened an R&D center in Lyon (bringing together 5 engineers) to develop new processes linked to AI technologies. and industrial computing. A relationship of equality between employee and manager “Beyond the fashion effect, digital transformation is also a real reality.

The sentence is final for companies that do not follow the march. This fourth revolution is different from the others: it is exponential. Each innovation goes further than the previous one. We are in a permanent technological break.

Digital transformation is a societal transformation driven by our uses,” explains Ilham Guggenheim, Managing Director of AKKA DS, which supports companies in their digital transformation. According to her, 52% of companies in the US top 500 (the equivalent of the CAC 40) have disappeared since the year 2000. The average age of a company was 67 in 1927. Today, this figure has fallen less than 10 years old.

New technologies have changed the game for both customers and employees. They have enabled the emergence of a new mobile customer who has the power and buys anytime, anywhere, on any device (ATAWAD/“Any Time, Any Where, Any Device”). As a result, the traditional marketing codes based on the 4 Ps (Product, Price, Place, Promotion) are obsolete and are replaced by the 4 Es (Emotion, Experience, Exclusivity, Commitment). At the same time, the corporate pyramid system has also become obsolete.

The employee becomes a creator of value, a source of wealth provided that he is better involved in the life of the company. According to the OECD, the lifespan of a technical skill was 22 years in 1970. It is only 2.9 years today. This development should lead to changes in our management methods. “Managerial innovation is one of the challenges. It is necessary to develop the employee experience (EX), that is to say the collaborator experience and to establish a relationship of equals between the collaborator and the manager. The key to success ultimately lies in combining the strengths of AI with those of employees.”

AI to rethink Lean. This is also the opinion of Cyril Dané, CEO of AIO, European leader in Karakuri Kaizen® and Lean manufacturing for whom: “Artificial intelligence without humans is nothing. The robot is still far from dethroning man. On the other hand, it opens up new avenues for improving well-being at work”. The entrepreneur continues by quoting Michael Ballé, co-founder of the Lean France Institute when he writes: "As Cecil Dijoux clearly shows in his latest book "Hyperlean in action", thinking digital requires thinking Lean to succeed in its scale. -up. Very fundamentally, the machines, whether it is an injection molding machine or an AI, will never know how to learn from the particular requests of the customers to adapt to the real demand and design a new improved version of them. themselves”.

This does not prevent Lean specialists from using AI to enable them to go beyond traditional models and rethink the Lean movement, like AIO which has just presented the new version in early 2019 of its Numii solution at the last CES in Las Vegas.

“Numii uses a revolutionary approach that goes beyond standard models. She breaks down movement into elementary gestures in order to have a better understanding and a disruptive approach to human movement and effort. The Numii database will be open to the world of research to continue long-term research in the medical field and to have a better idea of ​​what health for all and health at work is. A world without difficulty “Imagine in the factory of the future a world without difficulty, imagine the end of occupational diseases.

Numii is a swarm Intelligence, a swarm intelligence, a system of robots that collectively measures, collects and processes to generate the first database on human work” explains Cyril Dané. In parallel, AIO is also developing an intelligent connected glove called Hiifu (skin in Japanese), in order to measure the pressures and movements of the hands and fingers in an industrial environment. The idea is to build the largest database in the world on wrist and hand work, which accounts for 45% of MSDs (Musculoskeletal Disorders).

The use of collaborative robots as recommended by HumaRobotics, and/or mobile robots like those deployed by Supratec, reduces the hardship at work by reducing heavy and repetitive manipulations. It also meets the needs of manufacturers for whom space savings are important by providing more flexibility. “Our mobile robots offer the possibility of reconfiguring workshops without any civil engineering work. insists Arnaud Marche, technical director of Supratec. For Clément François, commercial manager of MakitLean, which offers innovative Lean layout solutions: no digitalization before integrating Lean approaches into the factory.

“Before moving to 4.0, you must first ensure that you have mastered 3.0” he says. It is still necessary to know how to surround yourself with the right people and call on the right experts: "Faced with the multitude of existing technological solutions, companies have difficulty in integrating all the changes that can affect the introduction of new technologies both at the business level and organizationally and humanly. To respond to this problem, we must adopt a global approach based on these 3 dimensions,” warns Thierry Lacombe, CEO of the SPC Group. As for him, Philippe Prigent, co-founder of Axsant, one of the precursors of Lean Manufacturing is formal: “Before transforming the company, it is necessary to assess the aptitude of the performance of the factory. Often people want to invest heavily in new technologies when if their tools were better used, they could achieve performance faster than if they invested heavily. Combined with Lean, AI and digital tools can thus enable the well-being and fulfillment of employees while improving performance.