Imagine a self-driving car that doesn't just stop suddenly, but can actually explain why it did so in a way humans understand. Instead of a 'black box' where decisions are mysterious, this work creates a set of rules and tools to make AI driving logic transparent. It's like giving the car a voice to justify its actions to the driver and the law.
AI in autonomous driving is often a 'black box,' making it hard for users to trust and for regulators to certify. This lack of transparency creates legal risks and slows down the social acceptance of self-driving technology.
A human-centric methodology for trustworthy AI and a set of XAI KPIs. They also produced multi-sensor datasets (camera, LiDAR, radar) and a cloud-based toolset for XAI development.
If you are a vehicle manufacturer dealing with the lack of social acceptance for autonomous features — this project developed a human-centric methodology that ensures AI decisions are explainable and ethical. This helps in gaining user trust and meeting safety standards.
If you are a certification body dealing with the difficulty of auditing AI driving decisions — this project developed a set of Key Performance Indicators (KPIs) on explainability. This allows for a standardized way to verify if an AI system is accountable and safe.
If you are a software provider dealing with biased or unpredictable AI behavior in traffic — this project developed XAI training processes and datasets. This ensures the AI is unbiased and its decision-making is transparent to the end user.
Based on available project data, no specific pricing is mentioned, but tools and datasets are intended to be made available via European OpenData and OpenTools initiatives.
The project focuses on scalable MLOps and a cloud platform complying with GAIA-X architecture to ensure the tools can be used in large-scale digital ecosystems.
Based on available project data, specific licensing terms are not listed, but the project emphasizes alignment with European Dataspaces for data sharing.
It addresses accountability, privacy, and ethics, providing a methodology that helps developers and legal bodies meet European values of data protection and trust.
The project runs from 2022-11-01 to 2025-10-31, with the core methodology already created in the first period.
The consortium is heavily weighted toward industrial application, with a 53% industry ratio comprising 9 companies, including 4 SMEs. With 17 partners across 7 European countries, the project balances academic research (3 universities, 3 research centers) with practical commercial implementation, suggesting a high likelihood of the results being applicable to real-world automotive products.
Contact VICOMTECH in Spain for methodology access.
Contact us to explore how to integrate XAI KPIs into your autonomous driving validation pipeline.
