•  2 minutes

Machine learning to support objects intelligence

Cartesiam, a start-up created in 2016 by Michel Rubino, Joël Rubino, François de Rochebouët and Marc Dupaquier, presented last February its NanoEdge™ AI Studio software, a revolution in terms of IoT (Internet of Things). This tool indeed makes it possible to turn object connectivity into true intelligence. Michel Rubino, CEO of Cartesiam, gives us a preview of this innovation.


Embedded AI software that is unique in the world.

NanoEdge™ AI Studio is an Integrated Development Environment (IDE) that comes in the form of a software. It can be installed on a Windows or Linux PC to make any device with a microcontroller intelligent. Power, vibration, sound, electromagnetic field or pressure etc. whatever the type of signal collected and processed by the sensor, this tool is efficient and reliable. “The software works like a stethoscope, except that there is also a doctor with it”. The algorithm analyses the ‘normal’ status of the device, then, thanks to automatic learning, it is able to detect deviations, according to the mathematical differences between what he learned initially and what he finds. It then sends an alert to warn of possible malfunctions. The microcontroller is programmed using the software. There are between 300 and 500 on a car. 15 billion ARM® Cortex® M microcontrollers are sold every year.

The strength of ‘Edge’: when local processing becomes a security asset

Until now, using AI and embedded devices required data scientists and using the cloud to collect and store the nominal data. The breakthrough with NanoEdge™ AI Studio is that it works locally. “Our strength is to embed learning and inference directly in microcontrollers. We process data precisely where signals become data. Most security issues disappear: users retain ownership and control over their data. It is a major benefit for all industry players, in particular”. This also generates cost savings as the company no longer needs rare and costly skills such as data scientists, thanks to the functions that will be integrated into the software which will allow the object to react alone without external intervention. The software also runs autonomously. If the Internet network collapses, the microcontroller keeps working, while using very little energy (one battery lasts more or less 3 years). 

An interesting prospect for predictive maintenance

NanoEdge™ AI Studio enables devices to self-diagnose. In a certain way, they become aware of their condition. In the future, an air conditioning system will for example be able to detect that its filter is clogged. The range of possibilities is therefore particularly vast, and numerous companies have already realised the interest of this solution, companies like Schneider Electric, Lacroix Electronics and Naval Group, that have already tested the software to maintain embedded equipment. “Nowadays, on the embedded AI market, Cartesiam is the only one to offer such an innovation. With NanoEdge™ AI Studio, traditional AI barriers have at last been removed. It becomes accessible to companies of all sizes that wish to design and sell innovative products easily, fast, and at a very competitive price.”