Imagine a machine that could detect a breakdown six months before it occurs. Science fiction? Not for Astriis. This Grenoble-based start-up, founded in 2022, has developed predictive maintenance technology that is revolutionising industry. Its four co-founders, among them Nicolas Saubin and Paul Roy (respectively graduates of Grenoble INP - Ense3 and Grenoble INP - Ensimag), have a clear ambition: to combine artificial intelligence and signal processing to predict faults before they become critical.
In the wind energy sector, an unexpected breakdown can immobilise a turbine for six months, while a faulty part is ordered and replaced. Thanks to Astriis, it is no longer necessary to wait for something to break down: the technology continuously analyses acoustic, vibrational and electrical signals in machines to detect the first signs of weakness. Scheduling an intervention at the right time reduces costs and environmental impact.
While other solutions provide only basic indicators or still require human intervention, Astriis pushes automation to an unprecedented level. “We have developed a signal processing method that is capable of identifying flaws with never-before-seen precision,” says Nicolas Saubin.
In three years, Astriis has capitalised on nine patents developed at GIPSA-lab and integrated them into its technology, established itself in France, South America and Japan, and acquired eight industrial clients in energy, metallurgy and papermaking. The company’s turnover reached €150,000 in 2024 and should exceed €500,000 this year.
But their ambition does not stop there. The start-up is getting ready to launch an industrial version of its SaaS platform in June, tested by EDF and TotalEnergies. This will allow companies to monitor their facilities online, without having to manage heavy infrastructure.
The recent certification by the MIAI Cluster marks a key step for Astriis. “This increases our credibility and opens new doors for us in the artificial intelligence ecosystem,” says Nicolas Saubin. Astriis sets itself apart by combining AI and signal processing, while most other companies concentrate on one or the other.
The start-up was selected for France 2030 and supported by ADEME and Bpifrance. It is also working on a pilot project for floating offshore wind turbines, technology which remains experimental but promising. To accelerate its development, particularly in Spain, Astriis is getting ready to raise funds and aims to develop its catalogue of detectable faults: 5 at present, 15 by the end of 2025 and 40 in the next three years.
* CNRS / UGA / Grenoble INP - UGA
