Europe to develop standards for game-changing AI

Europe is to forge ahead and develop standards to embed Artificial Intelligence (AI) in to future ATC operations and also pave the way for single pilot operations, writes Aimée Turner.

AI is a new domain where technologies combine the raw computing power of machines with the cognitive power to reason, learn and make decisions and is recognised as becoming increasingly important in the aviation sector.

AI technologies are developing fast and are becoming increasingly accessible. They provide attractive future capabilities thanks to a significant increase in processing power in recent years, allowing them to perform certain tasks as well as or better than a human.

The European Organisation for Civil Aviation Equipment has agreed to form a working group following this year’s EUROCAE Symposium and 56th General Assembly which took place in April.

Christian Schleifer-Heingaertner, secretary general of EUROCAE, tells Air Traffic Management that its EUROCAE membership wanted to initiate standards in this area. “AI is now something that our members feel is at a stage where, because they are using it increasingly in their products, they need a standard which might also serve in the future for certification by the European Aviation Safety Agency (EASA).”

The move is being driven by the leading European manufacturers of airborne equipment – although ground equipment manufacturers are understood to be looking at AI as well.

Schleifer-Heingaertner said that although procedures between the two are different which could eventually mean two different documents the EUROCAE working group will start working on one sole document. “We have standards for cyber protection in software and standards for AI may well develop in the same way – not the product but the process behind it,” he said.

The working group will start its work in summer and will develop an internal report within the first six months and the process standard within 36 months. The timescale will also depend on R&D and industrialisation activities by the manufacturing base which has already started to develop the first AI applications. “What they want now is to agree on an acceptable process in order to get AI solutions certified,” said Schleifer-Heingaertner.

He added that the process could possibly follow EUROCAE’s work on drone standards where the organisation was asked to come up with some simple standards that the emerging industry could use. “We took an approach that focussed on criticality and developed standards that offered guidance on that to new entrants into aviation who need to understand the aviation safety philosophy. AI may adopt the same approach although with more steps.”

At the recent EUROCAE Symposium Béatrice Pesquet, the ATM research and innovation director for Thales, delivered a presentation which stressed the need for harmonised regulation and standardisation. She cited the emergence of unmanned traffic management (UTM) which similarly required prompt effort to secure the appropriate level of global harmonisation in order to facilitate UAS integration. She said a defined AI standards framework was essential in order to ensure safety and security while enabling competitiveness and to define a path to the end concept with reasonable steps to facilitate the standardisation work.

“There is a strong need for the aviation community to ‘refresh’ industrial guidelines and standards to address the specific digital emerging issues of each aviation segment – both airborne and ground,” she said. She recommended a shift away from current legal regulatory requirements based on traditional ‘development assurance’ to a hybrid approach of both development assurance and learning assurance combined with an enhanced operational monitoring capability. “For Thales, AI is a game changer for the whole value chain – but the associated risk and challenges need to be well understood and addressed.”