Case study KIEZ4-0 – European certification of AI under industry 4.0
The aerospace industry is facing increasing complexity and limited capacity – yet current certification procedures are not designed for AI-based systems. As part of the KIEZ4-0 project, fortiss has developed formal verification methods and human-centred design strategies to make AI-supported aerospace applications certifiable. The result: fundamental tools and concepts that enable the reliable and demonstrably safe use of AI in one of the world’s most safety-critical industries.
The KIEZ4-0 project was part of the german aerospace research programme (LuFo VI-1) and was funded by the Federal Ministry for Economic Affairs and Energy (BMWE). The consortium comprised partners from industry, academia and certification bodies. Together with the European Union Aviation Safety Agency (EASA), work was carried out on future certification standards for AI in aerospace.
Challenge
AI systems offer great potential for the aerospace industry – for example, for autonomous or single-pilot operational concepts. However, under current safety standards, they are largely considered uncertifiable. Traditional aerospace certification processes are not equipped to deal with the complexity and opacity of modern AI technologies. There is a lack of robust methods for verifying and certifying AI-based systems.
Solution
fortiss focused on two key areas: formal verification and human-centered system design. Using methods such as model checking and theorem proving, the team demonstrated the reliability and safety of AI in flight control and assistance systems. In parallel, the fortiss team developed guidelines to address AI risks early in the development process and to ensure alignment between human expectations and AI behaviour. Demonstrators provided examples of how symbolic AI and formal proof methods can be applied to future cockpit assistance scenarios.
Result
- Development of formal verification methods to support certification of AI-based avionics functions.
- Guidelines for human-centered AI design to reduce AI-related risks and ensure operator alignment.
- Demonstration how symbolic AI can be integrated into certification-relevant workflows in aerospace.
- Built simulation-based proof-of-concept systems for flight diversion and mission assistance applications.
- Collaboration with EASA to ensure regulatory compatibility and advance harmonized certification procedures.
- Contribution to international standardization efforts for AI safety assurance in aerospace.
- Establishment of a framework that is transferable to other safety-critical sectors such as robotics and autonomous driving.
- Strengthening Germany’s role in shaping trusted AI integration into future air mobility systems.
Outcome
The KIEZ4-0 project demonstrates how formal verification methods and human-centred design are paving the way for the certifiability of AI in aerospace – including for innovative concepts such as single-pilot operations. This creates a solid foundation for the safe and traceable use of AI in highly critical applications and future aerospace systems.
Project partner
