The European Union Aviation Safety Agency published a report that addresses the challenges posed by the use of Neural Networks in aviation, compiled in conjunction with Daedalean AG, a Zurich-based start-up specialising in the processes required for future autonomous operation of aircraft.
The report is the output of a joint project, entitled “Concepts of Design Assurance for Neural Networks”, which examined the challenges posed by the use of Neural Networks in aviation within the broader context of allowing Machine Learning (ML) and more general Artificial Intelligence (AI) on board aircraft.
Experts from EASA and its partner Daedalean bundled their expertise to investigate the concepts and foundations of necessary safety guidelines for the application of Neural Networks in safety critical avionics.
From the EASA perspective, this report constitutes a first major step in the definition of the “Learning Assurance” process, which is a key building-block of the “AI trustworthiness framework” introduced in the EASA AI Roadmap 1.0. In particular the W-shaped assurance cycle developed in this report will serve as a key enabler for the certification and approval of machine learning applications in safety-critical applications.
The major challenges and risks associated with machine learning systems in safety-critical applications, as identified in the EASA AI Roadmap 1.0, were investigated during the project. They were addressed applying certain assumptions and based on a specific use case. The next step for EASA will be to generalise, abstract, and complement these initial guidelines, so as to outline a first set of applicable guidance for safety-critical machine learning applications.
Daedalean AG, founded in 2016, works with eVTOL companies and aerospace manufacturers to specify, build, test and certify a fully autonomous autopilot system. It has developed systems demonstrating crucial early capabilities on a path to certification for airworthiness.
