Artificial intelligence from every angle
Image recognition, chatbots, automated production chains… Artificial intelligence (AI) has become part of daily life, both in the home and on industrial sites. Daniel Duclos, Signal and Information Technology Department Manager at the Safran Tech research center, explains what it means for Safran.
Last July, Safran and seven other industrial companies signed a manifesto about the development of artificial intelligence in industry. What does AI mean for the Group?
We are exploring three major areas of application. In the first place, how AI affects our products. In the first place, this concerns equipment and defense and security systems, where AI can, for example, help with environment recognition in a conflict zone, and with the organization of military operations. We are also studying how this equipment can contribute to making ground and air vehicles more autonomous: automated aircraft taxiing to the parking area, autonomous navigation assistance for future 'flying taxis' or logistical drones, etc.
The second area of application: services. Here the challenge is to optimize the use, lifespan and holding cost of our equipment for our clients, in particular by improving fleet management and predictive equipment maintenance.
Finally, AI can enable us to improve the Group's efficiency: in production, by automating component control and by optimizing manufacturing procedures, as well as in engineering, by giving the design department simulation and modelling tools capable of performing more complicated calculations in real time, making more reliable predictions, and designing more and more complex systems…
Watch the video
Are some of this work going to come through soon?
The first applications should see the light of day in the Group's industrial activities next year. Achievements in defense programs should follow swiftly autonomous vehicles, intelligent ground robots, drones… Aeronautics will probably be the last sector to use AI due to safety requirements linked to international aeronautic certification standards.
What resources is the Group calling upon to make progress in the field of AI?
While there are many challenges, there are just as many ways to deal with them. To apply AI to critical functions such as those linked to flying an airplane, we need to develop design and validation methods so we can be fully confident in the machine's ability to make the right decisions and to ensure that it is 'reasoning' properly. Safran is dealing with these problems through partnerships with other companies, as is the case with the DEEL research project*. Safran is also contributing to the development of a standard and regulatory framework by participating in several workgroups in the aeronautics sector (like EASA** and the FAA***) and on normalization (such as AFNOR and ISO).
Several teams specialized in AI have already been recruited within the Group, and we are continuing to recruit specialists. However, to compete with the Web giants, which covet these same specialists, the industry has to make itself more attractive if it wants to obtain the talent it needs. It is the reason underlying our involvement in the Télécom ParisTech industrial chair dedicated to AI. The manifesto that was signed this summer is also intended to attract interest, from academics and start-ups, in AI applications for our business sectors, to help develop collaboration on research, and to attract innovation. Finally, an internal training cycle has been developed in conjunction with the engineering school CentraleSupélec.
Watch the video
Find out more
70 years of artificial intelligence
Work on AI first started in the 1950s. At the time, the idea was to make a machine capable of reasoning using a knowledge base and rules provided by people. A good example is the development of chess programs. Starting in the 1990s, and especially after 2010, the increase in computer calculating power meant researchers could design real "neural networks" inspired by the human brain, capable of learning and reasoning independently. We talk about "machine learning" and “deep learning", and one of the best-known applications is Apple's Siri virtual assistant.
* DEpendable & Explainable Learning. Research project launched in 2018 and conducted by the Saint-Exupéry Institute of Technology in Toulouse.
** European Aviation Safety Agency.
*** Federal Aviation Administration (United States).