The innovation cluster coordinates all our internal and external innovation activities, as well as the relations with our research ecosystem (customers, competitiveness clusters, universities, research laboratories, etc.)
The innovation cluster is headed by Frédéric Onado who, during his career at the ADVANS Group, held senior positions at ELSYS Design, AViSTO and a start-up company.
Since the creation of the innovation cluster, around a hundred engineers have been involved in our various projects (start-ups, PoCs, collaborative projects and customer projects).
The acquired skills are capitalized within our group and then shared, for example during technical events organized with our engineers within our various locations.
Frédéric, can you please introduce us one of the R&D areas of ELSYS Design?
In the field of the autonomous vehicle, but also of the smart city, one of the key challenges will be to realize the processing of the artificial intelligence no more on servers, but at the electronic board level, as close as possible to the sensors. Basing the operation of the autonomous vehicle on transmissions to servers then feedback is inefficient and very energy-intensive, not to mention the increased risks in terms of cyber security.
Our expertise in FPGAs, electronic boards and embedded software are our key assets to meet these challenges.
Which innovation project are you the most proud of?
Based on the ideas of our engineers, we founded a start-up in the field of powerline communication technology and positioned it in the worldwide top 5, against players like Qualcomm, Broadcom or Marvell. It is now backed by one of the world’s leading semiconductor companies.
At the same time, we have actively contributed to the standardization of this technology; I was myself a member of the steering committee of an organization that was at the origin of several international standards.
Last question: what do you think is our most promising project?
It’s hard to name one. The UPmem project is exciting: it consists in optimizing data processing architectures by offloading some of the main processor’s calculations to DRAM-integrated co-processors. This makes it possible to limit the exchange of data with the memory and thus to improve the overall efficiency of the system.