NOTE! This site uses cookies and similar technologies.

If you not change browser settings, you agree to it. Learn more

I understand

Learn more about cookies at :

October 2, 2020 | SACHEMS SHM development and testing platform

sachems 250
© Moovlab

​CEA List, a CEA Tech institute, is playing an active role in building the SHM (structural health monitoring) industry. The institute now offers development and testing support for SHM technologies at a new platform called SACHEMS.

​​In structural health monitoring (SHM), sensors are integrated into a structure to monitor its condition in real time, providing information that is useful for preventive maintenance, for example. For structures like aircraft, nuclear power plants, wind turbines, and bridges—for which inspection creates undesirable downtime or safety hazards—SHM is an ideal strategy. Although SHM research has been intensive over the last twenty years, solutions have only recently begun to reach the market, thanks mostly to advances in digital technology.

In 2020 CEA List, a member of the Carnot Network, created a new research and innovation platform on the CEA Paris-Saclay campus dedicated exclusively to SHM. The idea is to create a space where academic and industrial stakeholders, technology providers, and the end users of SHM solutions can come together.

SACHEMS, which stands for SAClay High-end Equipment for the Monitoring of Structures, is home to workstations with equipment specific to SHM. SHM solutions providers will be able to use the facility for their proof-of-concept prototyping and testing, benefiting from a shared library of mature enabling technologies and, as a result, shorter development cycles. Partners will have access to servers for their computing and database needs, as well as to the most advanced sensor technologies (piezoelectric, magnetostrictive, fiber optic) and sensor networking tools. The platform is also home to design tools (electronics design for embedded systems) and testing facilities (climatic chambers, electronics test benches).

CEA List hopes to bring not-yet-mature SHM technologies to a stage where they can be scaled up and commercialized on a variety of industrial markets. SACHEMS, made possible in part by financing from the Ile-de-France regional government, will help do exactly that.

Read article at 



September 24, 2020 | Trusted AI: new advances in the formal validation of neural networks

IAconfiance 250
© AdobeStock

Researchers from CEA List, a CEA Tech institute, have trialed a new approach to the formal validation of neural networks applied to image recognition. Their goal is to improve the safety of features such as pedestrian detection.

When it comes to artificial intelligence, proving that a neural network has been "taught" to do something like recognize pedestrians is a challenge. The formal validation of neural networks applied to image recognition, essential in critical use cases, has been a significant technical hurdle. Before a vehicle's ability to effectively avoid all pedestrians can be tested, for instance, it is important to unambiguously specify what constitutes a pedestrian.

Since it is impossible to describe a pedestrian in mathematical terms, the CEA List Carnot Institute researchers came up with the idea of using image generators (also known as simulators) to "train" neural networks. This avenue seems all the more promising given that simulators are now in widespread use, particularly in the automotive industry, to compensate for the lack of real-world training data. The new specification formalism developed places the simulator at the heart of the validation process, where it serves to formally specify the properties the neural network must satisfy. This allows conventional formal analysis methods, one of CEA List's areas of expertise, to be applied to the network to verify its compliance.

The theoretical results produced were included in the proceedings of the European Conference on Artificial Intelligence (ECAI 2020)[1], while proof-of-concept testing was completed on a scaled-down simulator specially designed for the purpose. The next stage of the process will involve validating the theory on a full-scale system, but this pioneering advance is a significant and tangible step towards trusted AI.

[1] Girard-Satabin, J.; Charpiat, G.; Chihani, Z. & Schoenauer, M. "CAMUS: A Framework to Build Formal Specifications for Deep Perception Systems Using Simulators" ECAI 2020

Read article at



September 16, 2020 | A decision-assistance tool for crisis management

expressif 250
© AdobeStock_Anton Gvozdikov

ExpressIF®, the artificial reasoning platform developed by CEA List, a CEA Tech institute, has been augmented with a spatial data processing module to assist in decision making. The technology is currently being used to manage wildfires, but it also offers potential in other emergency management situations.

Symbolic artificial intelligence algorithms, which simulate complex reasoning and apply it to heterogeneous and uncertain data, can be extremely useful decision-assistance tools in crisis management scenarios. As part of a project backed by the European Regional Development Fund (ERDF) and France's Occitanie regional CLE fund, CEA List, a member of the Carnot Network, partnered with CGX, a company specializing in geographical information systems, to bring the reasoning capabilities of ExpressIF® to wildfire management professionals.

New spatial data processing capabilities were implemented to more effectively assess a potential on-site response and help determine the best course of action. In the first module, the "rules" expressed in structured language (close to natural language) were used not only to model the spatial reasoning to be applied, but also to process both detailed geographic data and weather forecasts to predict the fire's development over time and space. The second module used the temporal reasoning capabilities of ExpressIF® to monitor the temperatures of assets like vehicles and people on the ground, assess the associated risks, and raise alerts in the event of danger.

The human-machine interface (HMI) developed by CEA Tech Occitanie provides a dedicated module for inputting data (locations, events, means, resources, etc.) and displays the results directly on a map in a web app. Further developments to track the rules applied are underway, with the goal of delivering an automated explanation of the results.

The unique spatial and temporal reasoning capabilities of ExpressIF®, combined with its ability to intelligibly process uncertain data, will be extended and applied to other critical scenarios, setting the platform on a course to become a leading AI technology for crisis management.

Read article at



August 18, 2020 | Predicting induced radioactivity in medical accelerators

radioact 250
© Adobe stock - Siarhei

CEA List, a CEA Tech institute, has mapped the parts that present induced radioactivity in medical particle accelerators, information that is vital to dismantling the equipment safely.

Increasing numbers of the particle accelerators used in radiation therapy are reaching the ends of their lifespans, which raises the issue of dismantling the machines and managing radioactive waste. Particle accelerators contain parts with complex geometries, some of which become radioactive due to their activation with accelerated beams and exposure to the secondary particles generated by interaction with radioactive material. The first challenge is knowing exactly which parts are radioactive. The types and amounts of radioactivity must also be determined. ANDRA, France's national radioactive waste management agency, turned to CEA List, a member of the Carnot Network, for help dealing with this new problem.

CEA List lab LNHB (Laboratoire National Henri Becquerel) utilized geometric data from the Varian radiotherapy accelerator at the DOSEO platform and leveraged CEA List's expertise in Monte Carlo simulations to model the functioning of a medical accelerator and its consequences on the parts that make up the machine. The researchers were able to map the induced radioactivity in the different parts and classify them by the type of radiation emitted (beta or gamma rays).

ANDRA will use the data to ensure that radioactive waste produced from the dismantling of medical accelerators is managed properly. This groundbreaking research positions CEA List as a source the international community can turn to for an accurate and reliable methodology for the management of radioactive waste from medical accelerators.

 Read article at 



August 4, 2020 | Papyrus keeps an eye on digital systems

papyrus 150

Of the many cybersecurity tools developed for industrial software users by CEA List, a CEA Tech institute, the open-source Papyrus platform is a key technology. Not only can Papyrus be used to guide the development of software that protects users’ personal data, but it can also model risk analyses.

In a world where digital technology is ubiquitous, the demand for powerful data protection and cybersecurity solutions will only increase. Therefore, the quality of data protection and cybersecurity specifications will become a decisive factor in the development of complex systems. CEA List, a member of the Carnot Network, has leveraged its expertise in model-driven engineering, through its Papyrus software development platform, for several projects.

CEA List researchers have developed a methodology and the associated tools to ensure that personal data protection regulations (GDPR) are factored in from the initial system design phases. The tool developed by CEA List in research for the EU PDP4E project allows legal obligations to be expressed as functional requirements and technical constraints, so that a system architecture can then be built in a way that guarantees that these obligations are met. The next step will be to test the tools on partner companies’ use cases to ensure that they are robust.

In research being conducted for the ModSecAéro project in partnership with the French Directorate General for Armament, CEA List researchers are developing a methodological framework to assess the resistance of embedded aeronautics systems to cyberattacks. The tools, which leverage the Papyrus platform, will be capable of modeling risk analyses utilizing a methodology that is compliant with the industry’s standards. With the tools being developed, critical components and the associated threats will be identified, the possible attacks on these vulnerable elements will be determined, and countermeasures will be implemented. These functions will be integrated into a model that will improve the analysis over time. A partnership between software developer Trialing and CEA List is expected to lead to the commercialization of the tools.

Read article at