AI & ML Firmware

The downtime of industrial machines, engines, or heavy equipment can lead to a direct loss of revenue. Accurate prediction of such failures using sensor data can prevent or reduce the downtime. With the availability of Internet of Things (IoT) technologies, it is possible to acquire the sensor data with high frequency. Machine Learning and Deep Learning (DL) algorithms can then be used to predict the part and equipment failures, given enough historical data. DL algorithms have shown significant advances in problems where progress has eluded the practitioners and researchers for several decades.

Sensors convert physical signals into electrical signals. This makes it possible to measure physical quantities in the environment. If such measurements are made repeatedly and stored, the behavior of the physical quantity can be studied to gain valuable insights.

ML offers algorithms that learn from data. The algorithms learn a representation of the training data, which is then used to make predictions on out-of-sample data.

Deep learning is a machine learning technique that teaches computers to do what comes naturally to humans: learn by example. Deep learning is a key technology behind driverless cars, enabling them to recognize a stop sign, or to distinguish a pedestrian from a lamppost. It is the key to voice control in consumer devices like phones, tablets, TVs, and hands-free speakers. Deep learning is getting lots of attention lately and for good reason. It’s achieving results that were not possible before.


The EDGE TECHNOLOGIES project aims to develop an “AI” based on deep neural networks which will equip the whole range of our future intelligent measurement sensors (in situ information processing), very low consumption, and which will have for vocation convert traditional industrial equipment (pipelines, valves, pumps, motors, etc.) into intelligent devices connected wirelessly. We are particularly targeting the sectors related to transformation and energy processes (petrochemical industry, bioenergy, storage, etc.) which are seeking disruptive and secure solutions. The sensors we develop measure the key parameters of the equipment’s surface (vibrations, noise, temperature, etc.) which are analysed automatically (on-board AI algorithms), on site and in real time in order to obtain significant information on its condition and its performance, thus allowing users to identify inefficiencies in their system and reduce the risks associated with operation and maintenance. As a result, maintenance can thus be planned according to actual needs rather than on the basis of generic schedules. This extends the life of the equipment, reduces maintenance costs and reduces or prevents unexpected downtime due to breakdowns while limiting the risk of access to hazardous industrial sites.


  • For this purpose, we are collaborating with IN2MP laboratory for their expertise in embedded AI in microchips.
  • Located in Marseille and Toulon, IM2NP is a UMR (7334) under triple supervision of the CNRS, AMU and the University of Toulon (UTLN).
  • 19 research teams gathered in 5 scientific departments.
  • Over 2,500 publications.