For the Gartner Group, the “Edge AI” will quickly become the standard thanks to a decisive advantage: via the Machine Learning functionalities embedded in a microcontroller, manufacturers can analyze and interpret data locally. This allows them to eliminate latency in terms of decision making and can act quickly and securely because the data remains on the periphery and does not need to be transmitted to the cloud. Processing and analysis are performed in a very energy-efficient microcontroller, whereas architectures that use neural networks pre-trained on servers are still very energy-intensive.
As Gartner explains in its report, “The 3 Barriers to AI Adoption”, AI skills and data availability are two of the three barriers to entry slowing down the adoption of AI and its mass deployment in enterprises. In a traditional model, data-set and data scientists are inseparable for the success of any project. However, the constitution of a set of relevant data and the availability of data scientists to work on this data and develop algorithms based on deep learning (Deep Neural Network) are the sine qua non condition for the development of solutions.
Thus, industrialists no longer have to resort to these increasingly rare expert profiles and data sets that are difficult to build up to develop machine learning solutions in the “Edge.” Semiconductor manufacturers have understood this. They are investing more and more in this type of technology, and several of them have formed partnerships with Cartesiam. They enable their customers to keep control of their data by having it analyzed directly at the source (Edge), where the signal is transformed into data, rather than in the cloud.
The signals analyzed at the source can be diverse and varied. It can be the intelligent wind turbine assistant, learning and then monitoring the vibrations of a machine as Bob does. The aim is to detect anomalies and carry out predictive maintenance operations as soon as necessary. This solution, which is based on NanoEdge™ AI from Cartesiam, is already being massively deployed in European companies of all sizes and, more recently, in the USA. Other Cartesiam customers, both design offices and end customers, are working on solutions based on NanoEdge™ AI but using different signals such as sound, magnetometry, pressure or air pollutants.
Every year, 40 billion microcontrollers are sold, 15 billion of which are connected to the world economy. These components are integrated in all everyday objects, from coffee machines to air conditioners to cars and, of course, in all connected objects. It is in this embedded AI market that Cartesiam aims to become the world leader.
“Being identified in the Gartner Hype Cycle 2019 as one of the 3 suppliers of the Edge AI brick is a great honour and recognition for the entire Cartesiam team. Our technology is the result of several years of research and development where we chose to rewrite, from algebra, all the machine learning algorithms so that they can run in an environment as frugal as that of a microcontroller. “commented Joël Rubino, CEO and co-founder of Cartesiam.
He adds, “That’s how we created NanoEdge™ AI, a next-generation Machine Learning engine. With our technology, we enable design offices and manufacturers to easily integrate Machine Learning functions into all everyday objects. We are several years ahead of our global competitors. This explains the speed of our international development and the recognition of the Gartner Group.
Founded in 2016 and headquartered in Toulon with offices in Paris and New York (USA), Cartesiam is a French software publisher whose mission is to develop artificial intelligence solutions on microcontrollers. NanoEdge™ AI, an innovation developed by Cartesiam, provides cognitive functions (Machine Learning) to microcontrollers embedded in connected objects. Whatever their industry, Cartesiam thus brings immeasurable benefits to its customers involved in projects integrating microcontrollers: Simplicity of deployment, secure environment, rich analysis, and reduced energy expenditure. The company has around twenty employees. The company, co-founded by Marc Dupaquier, François de Rochebouët, Michel and Joël Rubino, is located in Toulon for the R&D and in Paris and New York for the sales and marketing.