Artificial intelligence and machine learning are often thought to be too far off to have application today, but the Army’s Training and Doctrine Command’s intelligence unit recently tried to identify areas where such technologies could be useful for the force today.

One specific area they found was airspace deconfliction. In other words, preventing two aircraft from flying in the same area.

This is “a recognized problem across aviation, across flyers [and] across maneuver,” Chip Retzlaff, intelligence chief of information management at TRADOC, said March 26 during a presentation at the AUSA Global Force Symposium in Huntsville, Alabama.

TRADOC leaders pulled together requirements to see if they could facilitate airspace management and airspace deconfliction in contested environments.

In a video, Retzlaff showed that airspace deconfliction is planned manually. Workers have to manage airspace between a variety of manned and unmanned assets to include fixed and rotary wing aircraft, unmanned aerial systems and even missiles, all while maximizing each platform or weapons’ capability.

Artificial intelligence can help automate this process by automatically adjusting and recalculating when conditions change.

One example the video showed is if an Apache helicopter leaves its dedicated space to avoid ground fire or to kill a high value target, artificial intelligence could hold all friendly fire until the Apache is safe. Such a system could also alert the Apache pilots when the helicopter left its flight path.

In addition, such a system could also make changes depending on what’s important to ground forces at that moment. This could include opening airspace for attack aircraft in the midst of an enemy counterattack.

“What we’re trying to do is raise that awareness and then pull the community of folks that handle the requirements, who have the research and development folks together to get it addressed as a hard problem,” Retzlaff said

Problem of data

Still, one of the challenges with AI and machine learning space today is the need for data to train the algorithms.

Retzlaff explained Army workers are searching for a rich operational environment to extract data from to train algorithms to be used in other areas of interest. This could include intelligence preparation of the battlefield, wargaming, decision support tools and after action review visualizations.

He said officials realized such data may be available from the National Training Center, in which units go to simulated conflicts against near peer adversaries.