TrustME

The Lufo 7.1 project TrustME addresses the industrial standard for trustworthy AI in Airbus production environment. Using certifiable methods and practical tools, we make AI applications traceable, safe, and applicable. The sovereign AI stack combines decentralised innovation with central reliability and security within a federated system of sites and domains.

Airbus Operations leads the consortium with a focus on establishing the overall certification framework and coordinating the integration of AI-driven autonomous systems for the final assembly and equipping of major aircraft components. 

Airbus Aerostructures GmbH (ASA) focuses on designing a scalable AI stack for production and a harmonized AI-Governance model to enable self-learning process optimisation and automated quality assurance within the production of assemblies and large-scale components.
 

IMOD Validation Platform Picture

IMOD Validation Platform Image

The primary hurdle for AI in aviation is the lack of certifiability within manufacturing and production. Decisions made by AI are often perceived as "black-box systems," lacking the transparency and traceability required for safety-critical industrial deployment. Existing EASA guidelines focus mainly on airborne systems, leaving a research gap for production facilities. TrustME aims to close this gap by developing architectures and procedures that ensure AI in manufacturing is transparent, verifiable, and safe.

TrustME establishes a Foundation for certifiable AI through a structured V&V (Verification and Validation) approach. Implementation is divided into four pillars:
 

1. System Architectures: Defining a holistic architecture based on the TOGAF standard to make AI applications certifiable.
2. AI Platform & Data: Building a scalable, secure platform for training, testing, and managing AI models with high data quality.
3. Certification Methodology: Establishing principles for "Trustworthy AI" in alignment with the EU AI Act and EASA AI Roadmap.
4. Laboratory Validation: Testing AI models under production relevant condition in laboratory environment, such as fuselage and component assembly, to prove technical feasibility and compliance.

The TrustME project provides the rulebook and the basic software stack for certifiable AI. 
This results in significant industrial advantages

• Increased Efficiency: Significantly reducing lead times,  quality and resilient production systems.
• Safety & Trust: Enabling the use of AI in critical processes through transparent, human-traceable decision-making.
• Sustainability: Reducing scrap rates and material waste through intelligent process stability and predictive maintenance.