Carnegie Mellon University reportedly develops innovative version of AI pilot. The advancement is hoped to allow the autonomous pilot to navigate crowded airspace in future.
Possible applications of these types of aircraft range from crop protection and infrastructure inspection to goods transportation and air taxis.
Similar technology can be observed on commercial aircraft that fly at higher altitudes. However, this does not easily translate to lower-altitude airspace that utilises visual flight rules (VFR) rather than instrument flight rules (IFR). This being the primary area the technology is hoped to be used for it is vital for it to have the ability to function around piloted aircraft.
The team’s goal is for the robot to interact with other aircraft in the same way a human pilot would. Six cameras and a computer vision system allow the AI to ‘see’ other aircraft while automatic speech recognition using natural language processing provides speech enabled communication capabilities.
Training the AI pilot with data from airports, Allegheny County and Pittsburgh-Butler airports, the robot has experienced air traffic patterns, aircraft images and radio transmissions. Subsequently, the AI undergoes testing with flight simulators incorporating both the pilot and a human pilot. According to their results it can successfully navigate piloted aircraft, independent of the experience level of the pilot.
