Researchers have developed a unique artificial intelligence system that has the ability to pilot crowded airspace on its own, without the need for human pilots.
Much has been said about A.I. in the past few years. Different artificial intelligence systems have been developed, all serving specific objectives.
Not long ago, we wrote about LaMDA, Google’s state-of-the-art A.I., and a former Google engineer who had claimed that the A.I. was sentient. While this isn’t the case –likely– it shows just how far the development of artificial intelligence systems has become.
In the scientific community, for example, artificial intelligence is widely used. On the one hand, astronomers use artificial intelligence to explore the cosmos. But on the other hand, similar systems can be used to go through astronomical data and make discoveries that scientists may not have seen in the first place.
Now, researchers may have developed another, entirely unique A.I., one that could make pilotless aircraft a reality.
An A.I. Pilot
The Carnegie Mellon University team claims to have created the world’s first artificial intelligence pilot, enabling autonomous aircraft to fly in crowded airspaces.
The research was supported by the U.S. Army Research Office and the Army Futures Command’s Artificial Intelligence Integration Center (AI2C).
Air traffic controllers and pilots can communicate on the radio with artificial intelligence and avoid collisions. It can also predict other aircraft’s intent, track and coordinate their actions, and communicate with their copilots. Eventually, the researchers hope to make their A.I. so similar to a human pilot’s that it cannot be distinguished from one.
“We believe we could eventually pass the Turing Test,” explained Jean Oh, an associate research professor at CMU’s Robotics Institute (R.I.) and a member of the A.I. pilot team, referring to the test of an A.I.’s ability to exhibit intelligent behavior equivalent to a human.
Artificial intelligence communicates its intent to other aircraft, whether they are piloted or not, using both vision and natural language. By following this behavior, navigation becomes safer and more socially acceptable.
Air traffic patterns, images of aircraft, and radio transmissions collected at Allegheny County Airport and Pittsburgh-Butler Regional Airport were used to realize implicit coordination.
A computer vision system and six cameras help the A.I. detect nearby aircraft in a manner similar to human pilots. In addition, with its automatic speech recognition feature, pilots and air traffic controllers can understand incoming radio messages and communicate with Artificial intelligence through speech.
With the advancement of autonomous aircraft, drones, air taxis, helicopters, and other aircraft will be able to perform many tasks without a pilot behind them — moving people and goods, inspecting infrastructure, treating crops to protect them, and watching out for poaching and deforestation.
Nonetheless, the small aircraft, medical helicopters, and other aircraft will be flying in an already congested airspace.
It has been proposed that the FAA and NASA divide this urban airspace into lanes or corridors, limiting the types and number of aircraft that can use each. It would significantly alter current practices and usage in this airspace, creating air traffic jams that would hinder critical aircraft, like medivac helicopters, from reaching their destinations.
Airliners and other aircraft flying above higher altitudes under instrument flight rules (IFR) often use autopilot controls. Still, it has been difficult for the aerospace industry to develop an A.I. to handle the often crowded and pilot-controlled lower-altitude traffic flying under visual flight rules (VFR).
An A.I. designed by the team interacts seamlessly with aircraft operating in VFR airspace.
“This is the first A.I. pilot that works in the current airspace,” revealed Sebastian Scherer, an associate research professor in the R.I. and a member of the team. “I don’t see that airspace changing for UAVs. The UAVs will have to change for the airspace.”
Although the A.I. pilot has not been tested on actual aircraft, it has performed well on flight simulators. Using two flight simulators, the team tests the A.I. A.I. controls one, and humans control the other. Airspace is shared by both. Regardless of whether the pilot behind the controls is an experienced pilot, A.I. can safely navigate the piloted aircraft.
Artificial intelligence could be used to help autonomous aircraft carry passengers and deliver packages commercially. Ideally, drones and air taxis would not need a pilot to save weight and avoid pilot shortages.
“We need more pilots, and A.I. can help,” said Jay Patrikar, a Ph.D. student in the R.I. who worked on the project.
Join the discussion and participate in awesome giveaways in our mobile Telegram group. Join Curiosmos on Telegram Today. t.me/Curiosmos
