ICAO has teamed with Flight Service Bureau to develop the first application of AI technology in the field of aeronautical information management – AIM, writes Aimée Turner in the latest issue of Air Traffic Management magazine.
The NOTAM Organisational and Recognition Model – or NORM for short – has been developed to help assess the criticality of NOTAMs.
NOTAMS are issued whenever an important piece of information is needed to be made available to the aviation community. However, at any point in time, there may be more than 30,000 active NOTAMS in the system.
More than one million NOTAM messages are issued per year, including irrelevant or barely readable jargon and it has become a serious and time-consuming challenge to distinguish the highly critical NOTAMS from the less important items as the number of NOTAMs skyrockets on the back of rapid industry growth.
ICAO, concerned about the potential for human error, started looking at smart, cost-effective solutions using AI and Machine Learning technology which users could easily access.
Together with Flight Service Bureau which monitors international airspace and airports for changes that affect pilots, dispatchers, and aircraft operators, the aviation standards organisation built the AI bot supported by the latest Deep Learning technology.
“We gathered more than 17,000 responses from air traffic controllers, dispatchers and pilots about how they would score about each NOTAM messages and trained NORM on the collected responses,” said Marco Merens, ICAO chief of integrated aviation analysis. “Now NORM understands the context of each NOTAM and classifies the criticality score from 1 (not significant) to 5 (very critical), just like humans do.”
NORM is now accessible through the web at ow.ly/MLmZ30lGdJC where users can easily find out which NOTAMs are worth attention.
ICAO said it will further develop and promote analysis solutions such as NORM through the iSTARS platform and will continue to actively explore how AI technologies can assist the data-driven decision-making process.