Reducing time-to-market while ensuring reliability

Reducing time-to-market while ensuring reliability

 

From need to smart testing — towards more reliable products designed faster

Un collaborateur de SII Research sur son ordinateur

From need to product: reliable and high-performance digital solutions

SII explores digital solutions that accelerate the transition from requirement to product, while ensuring a high level of reliability and compliance. Our research approach focuses primarily on the following objectives:

  • Automate validation and verification through intelligent testing.

  • Leverage AI in all phases of the product development cycle—from requirement capture to production.

  • Accelerate the design cycle through MBSE, simulation and digital twins.

OUR RESEARCH TOPICS

une main écrit sur le tableau

Sketch to code/CRAFT

This project explores a new way of designing web applications using generative artificial intelligence. The first phase involves transforming a simple sketch made on a whiteboard into a functional web interface, with automatic HTML code generation and successive iterations in interaction with an LLM (or Large Language Model).

The second phase extends this approach to the complete prototyping of a software chain: UX/UI mock-up, application architecture, and code. It incorporates the choice of a framework, a graphic charter, and several technical or functional options.

The process is based on continuous improvement loops, focusing on accelerating design while improving the quality of prototypes.

 

 

Discover the project Sketch2Code
Ecran d'ordinateur de data

Validation Assistant

Non-regression testing is central to validating the evolution of simulation packages. Without it, we cannot guarantee the quality of pilot training on simulators or the robustness of tests on different systems on test benches. However, this task can quickly become extremely time-consuming. The aim of VA (Validation Assistant) is to reduce this time and allow the engineer in charge of regression testing to focus on the why rather than on the comparative study of the flight parameters of the simulated aircraft.

To do this, we use a machine learning approach based on data that has already been evaluated by engineers in the past. This enables us to reduce the analysis time for non-regression testing from five days to just one day. This project, which originated in our R&D department, was first presented at the internal Innovation Challenge and then funded through a CORAC (state funding from the DGAC). Today, this project has been industrialised at our client Airbus and is regularly maintained to ensure continuity and the inclusion of new aircraft simulations.

Bureau de l'agence de Lyon

MBSE For Test Means

Carried out in partnership with Airbus, the MBSE (Model-Based Systems Engineering) For Test Means project has significantly reduced test facility development times by creating digital continuity between the teams designing the systems and those developing the test benches.
Our solution, which we have industrialised for our customers, has automated many low value-added tasks and facilitated the configuration management of test equipment.

Ecran d'ordinateur de data

Assistant Systems Engineers (SIRIUS, CONDOR)

This project aims to enhance the reliability and speed up the production and validation of system engineering documents drafted from specifications. It relies on the automatic analysis of natural language documents to verify their compliance with requirements, a task that is currently mostly manual and prone to errors. The solution detects ambiguities, inconsistencies, and incompleteness in relation to requirements writing standards. It also generates compliance reports and associated SysMLv2 models and representations. The goal is to reduce costs, improve the quality of specifications, and secure the design of complex systems.