The Department of Defense is looking for AI-enhanced target recognition to help troops, vehicles and ships destroy drones.
The C-UAS Close-In Kinetic Defeat Enhancement project focuses on aided target recognition, or AiTR. This uses concepts such as AI, machine learning and computer vision to create a system that can detect threats — and distinguish them from non-threats such as birds — faster than a human operator can.
The first phase of the project is aimed at remote weapons stations, and specifically the ubiquitous Common Remotely Operated Weapon Station, or CROWS, turrets fitted to a variety of military vehicles.
“The primary objective is to accelerate the engagement timeline, initially focusing on Unmanned Aircraft Systems (UAS), with a secondary focus on other threats like vehicular and man-sized targets,” explained the Defense Innovation Unit solicitation. The deadline is May 15.
Prototypes must “demonstrably improve” the ability of current remote weapons stations to detect, track and engage Groups 1 and 2 — targets with a weight of 55 pounds and under.
Detection should be at ranges greater than 600 meters, and engagement at a minimum of 100 meters. The system should be effective against drones moving at speeds of at least 30 meters per second, or 67 miles per hour, per the solicitation.
The second phase of the project seeks to boost C-UAS capabilities on “both moving and stationary platforms, including ground and maritime environments,” the solicitation said.
Specifications include the ability to hit a Group 1 drone — under 20 pounds — moving at 7 meters per second, or 16 miles per hour, at a range of 50 to 200 meters. Weapons should be able to engage targets while depressed to minus 10 degrees or elevated directly overhead to 90 degrees, the document stated.
This requires contractors to provide a prototype that can “be fired in land and maritime environments,” the solicitation says, “rather than just a laboratory setting at time of pitch.”
Most noteworthy, meanwhile, is the third phase of the project: adding aided target recognition to small arms carried by dismounted troops.
“Desired solutions include systems capable of deflecting or self-aiming standard-issue rounds to increase hit probability against manually selected, transient targets, while integrating networked sensor and small arms fire control systems,” DIU said.
The system must be capable of engaging drones moving at least 7 meters per second, and “must be adaptable to dismounted legacy small arms, scalable across calibers and configurations, and maintain baseline weapon performance in the event of system degradation or failure,” the document states. “A semi-automatic, live-fire capable prototype is required.”
The final phase of the project seeks to improve integration between sensors and weapons.
“A commercial wireless edge network architecture that bridges to military systems and the reverse is essential across all stages of this effort to manage data transfer from sensors and weapon/fire control systems,” the DIU wrote.
The U.S. military is beginning to embrace aided target recognition. The Army is already testing small UAVs equipped with AiTR to help infantry squads control drones.
But the Pentagon is also aware that AI and targeting is controversial.
The DIU project specifies that there must be a human in the loop. Solutions must strictly adhere to DOD’s AI Ethical Principles. Non-compliance “will result in immediate disqualification,” DIU warned.
Michael Peck is a correspondent for Defense News and a columnist for the Center for European Policy Analysis. He holds an M.A. in political science from Rutgers University. Find him on X at @Mipeck1. His email is mikedefense1@gmail.com.