SAFEXPLAIN

SAFEXPLAIN is enabling the use of AI in safety-critical system by closing the gap between AI capabilities and functional safety requirements.

About SAFEXPLAIN

Began in 2022 and set to end in September 2025 with six partners involved, SAFEXPLAIN’s ambitions have been:

Architecting DL solutions enabling certification/qualification.
Making them explainable and traceable and adhere to “safety culture” through extension of traditional process lifecycle.
Preserving high performance.
Tailoring solutions to varying safety requirements.

Main Activities

Its key activities include:

Rethinking the Safety Lifecycle: AI-Functional Safety Management Methodology.
Rethinking AI software: Move from “black box” to “gray box” with explainable AI.
Combining technologies while preserving performance and accuracy: SAFEXPLAIN platform.
Assessing findings in 3 key domains: Space, automotive, rail.

exida development’s Results

With the end of the project, exida development now has two primary owned results:

The V&V SCENARIOS

This is exida development’s open result.

The purpose of the Scenario Catalogue and Triggering Conditions is to define a set of scenarios to describe how the intended functionality works when a specific relevant situation occurs and provide an overview of functional insufficiencies and triggering conditions defined for each of the SAFEXPLAIN use-case domains: Automotive, Railway and Aerospace.

This result is open content and can be publicly accessed at this link:
https://zenodo.org/records/17175735

The V&V TESTCASES

This is exida development’s proprietary result.

In the context of AI safety critical systems, a number of test cases have been derived from the Scenario Catalogues and Triggering Conditions for each of the SAFEXPLAIN use-case domains: Automotive, Railway and Aerospace. Each identified test case includes the relevant information (pre-conditions, steps and acceptance criteria) to allow their execution and evaluation.

For more information and availability, please fill out the info request at the end of the page, and we will contact you as soon as possible.

Other European Projects

EdgeAI-Trust

EdgeAI-Trust is dedicated to developing an advanced, trustworthy edge AI ecosystem through cutting-edge hardware, software, and tools. The project aims to enhance decentralized Edge-AI operations that are secure, reliable, and sustainable.

https://www.edgeai-trust.eu/

SHAS-AI

SHAS-AI targets the HW/SW security and Al-based high-risk systems intersection, aiming to enhance the security, resilience, automated testing, and continuous assessment of Al systems.