Xwing, a US supplier of modular autonomy technology for aviation, and Daedalean, a Switzerland-based developer of safety-critical and certifiable Artificial Intelligence systems for situational awareness and flight control, are to share data, knowledge and processes in the field of machine learning and AI to accelerate time-to-market and enhance safety. They are working closely with the FAA and EASA to certify their software.
No machine learning based safety-critical applications have made their way into the cockpit of civil aviation so far; this requires a paradigm shift in how software and hardware have traditionally been assessed and certified as safe. Aviation regulators have been working on how machine learning fits into their design assurance processes and how to adjust the respective standards. Both companies believe that forming a consensus on their assurance approaches and sharing information is the best way to advance the development of certification standards that will allow an increase in safety more effectively than if they worked separately.
"In this emerging industry, it's as crucial to collaborate with fellow pioneers as it is to partner with regulators around the world," says Luuk van Dijk, co-founder and CEO of Daedalean. "With this shared undertaking, we will be able to demonstrate that increasing safety is driving innovation and that a collaborative approach to harmonise regulations and standards ensures that best practices are universally adopted."
Both Xwing and Daedalean have published papers that outline their approaches to certification; Xwing's Formal and Practical Elements for the Certification of Machine Learning Systems and Daedalean's Concepts of Design Assurance for Neural Networks series.
"At Xwing, we balance our commitment to a strong safety culture with our push for technical innovation," says president and CTO Maxime Gariel. "Our collaboration with Daedalean underscores this philosophy and the importance we place on sharing data, knowledge and processes to inform a credible path forward toward certification for the industry as a whole as we work closely with regulators."
While Daedalean and Xwing have independently developed frameworks for the certification of AI-based systems, both companies agree that higher levels of safety can be achieved on a shorter timeline by sharing knowledge and relevant flight data so that all stakeholders, including industry, regulators and the public, can make significant progress on the path to certifying AI systems.