DARPA's RACER Program Demonstrates Autonomous Ground Vehicles for Target Hunting in Fort Knox Exercise

In a significant step toward integrating robotic systems into future battlefields, the Defense Advanced Research Projects Agency (DARPA) conducted a demonstration on October 15, 2026, at Fort Knox, Kentucky, showcasing four autonomous unmanned ground vehicles (UGVs) developed under its Robotic Autonomy in Complex Environments with Resiliency (RACER) program. The exercise, requested by U.S. European Command (EUCOM), tested the vehicles' ability to navigate complex off-road terrain, detect simulated enemy targets using satellite imagery and onboard sensors, coordinate movements, and support simulated strikes—all while maintaining strict human oversight for any lethal decisions.[1][2]

One vehicle advanced independently up a hill, scanning the terrain before halting upon detecting a potential target. Two supporting vehicles then moved forward to provide cover, followed by a simulated rocket strike on a target approximately 300 yards away. The demonstration highlighted the first end-to-end integration from space-based imagery to tactical ground operations, with vehicles from performers like Overland AI integrated with Anduril one-way attack drones.[3][4]

DARPA technical advisor Dr. Jason Gregory emphasized that while mobility, navigation, and detection are autonomous, "they don’t pull the trigger until a human says pull the trigger," aligning with Department of Defense policy on human control of lethal force. The terrain at Fort Knox was chosen to simulate European operational environments.[5][6]

This builds on RACER's multi-year focus on resilient off-road autonomy and comes amid broader U.S. Army efforts to deploy unmanned systems for high-risk missions like reconnaissance, logistics, and casualty evacuation, reducing exposure for troops in contested areas such as potential European theaters.