FAA Picks SMART Airspace AI

The Federal Aviation Administration has picked Palantir, Thales, and Airspace Intelligence to build software that spots flight conflicts up to two hours early. (theaircurrent.com) Transportation Secretary Sean Duffy disclosed the effort on April 17 at a Semafor event, and industry reports say FAA Administrator Bryan Bedford is overseeing it inside the agency. The system is called SMART and is being developed as a predictive tool for the National Airspace System. (semafor.com, theaircurrent.com) Air traffic control today often works on a short planning horizon measured in minutes. Duffy said the new software would warn controllers to change a route “an hour and a half or two hours before the conflict even happens,” compared with roughly a 15-minute look-ahead now. (dronexl.co, theaircurrent.com) The basic idea is simple: build a live model of the sky that can project where aircraft, weather, runway limits, and traffic bottlenecks will be later, not just now. Airspace Intelligence says its platform uses a four-dimensional forecast — latitude, longitude, altitude, and time — to predict demand-capacity problems before they hit the system. (airspace-intelligence.com, faa.gov) That matters in a network that handles more than 45,000 U.S. flights a day and still depends on aging infrastructure the government is trying to modernize. FAA planning documents describe the National Airspace System as the world’s most complex aerospace system and call for new air traffic and airspace management tools. (aol.com, faa.gov) The near-term target is manned aviation, but the same kind of prediction is central to drone flights that go beyond the pilot’s direct line of sight. Thales said in March that the FAA had granted access to unfiltered national airspace radar data for North Dakota’s Vantis network, a system built to support beyond visual line of sight unmanned aircraft operations. (thalesgroup.com) That gives all three companies a distinct angle. Palantir brings large-scale government software, Thales has decades in air traffic systems and recent FAA-linked drone infrastructure work, and Airspace Intelligence already markets an artificial intelligence traffic platform that recommends flow-management actions. (thalesgroup.com, airspace-intelligence.com, bloomberg.com) The FAA has not publicly released a full contract package or technical architecture yet, and the companies are still competing rather than operating a finished national system. DOT and FAA officials are expected to provide more detail at a press event on April 21. (bloomberg.com, dronexl.co) If SMART works as described, controllers would get earlier warnings before congestion hardens into delays, and drone operators would be planning into a sky that is modeled further ahead. For the FAA, the project turns air traffic management into a prediction problem as much as a tracking one. (theaircurrent.com, airspace-intelligence.com)