Proponents of flying cars like to state how much less likely collisions would be up in the air, where everyone wouldn't be traveling on the same level, yet mid-air collisions between aircraft do already occur. Although certainly not as common as automobile collisions, approximately 10 to 12 aircraft do fly into each other every year, with many more reporting near-misses. This has led to the U.S. Federal Aviation Administration (FAA) mandating that by 2020, all commercial aircraft (and small aircraft flying near airports) must be equipped with a GPS tracking system, which would give more accurate information on their location than is provided by ground-based radar. Scientists from the Massachusetts Institute of Technology (MIT) has been tasked with creating an algorithm, that would use that GPS data to keep the planes out of each other's way.
Using six month's worth of data from San Francisco-area airports, a team led by Maxime Gariel, a postdoc in MIT's International Center for Air Transportation, created a computer system that places virtual horizontal hockey puck-shaped volumes of airspace around aircraft. These pucks represent where an aircraft is likely to go. If the edges of two pucks overlap, the pilots are warned that they're getting too close to one another.
The size of the pucks vary, according to how risky the situation is. Two planes flying parallel in the same direction, for instance, will have relatively small pucks. Once one or both of those planes start to move toward the other, however, their pucks will get larger, in order to provide a warning that much sooner.
To keep pilots from getting tired of receiving too many warnings, the system issues two types: a moderate alert, that lets them know that their current trajectory is intersecting that of another aircraft, and a high alert, that tells them a collision is imminent. Gariel and his team were concerned about the possibility of there being too many false alarms, so the puck algorithm was tested in a computer model, which incorporated air traffic data gathered over eight months from all the aviation radar systems in the U.S. The algorithm reportedly worked well, with a low false alarm rate.
The air traffic model, created at MIT's Lincoln Laboratory, did however have one limitation. All of its data was based on radar, which small aircraft often fly beneath. They particularly do so near airports, which is where 60 percent of midair collisions occur. In response, Gariel's team is now working on another model, that simulates the flight paths that small aircraft tend to follow around airports. They are also looking towards testing the algorithm on real planes.
See the stories that matter in your inbox every morning