U.S. Army seeks AI systems to automatically detect mines and IEDs

The U.S. Army has issued a formal request for information seeking industry solutions capable of automatically detecting explosive hazards and complex battlefield obstacles using artificial intelligence, signaling a push to reduce the burden on combat engineers during some of the most dangerous missions in modern warfare.

The solicitation, published April 7, 2026, by the Army Contracting Command–New Jersey at Picatinny Arsenal, invites defense companies to submit technical white papers describing automated target recognition systems capable of identifying anti-tank mines, anti-personnel mines, improvised explosive devices, unexploded ordnance, and man-made obstacles such as concertina wire and tank ditches. The response deadline is May 15, 2026.

The effort is led by the Army Capability Program Executive for Ammunition and Energetics, working alongside the Combat Capabilities Development Command Armaments Center and the DEVCOM Command, Control, Computers, Communications, Cyber, Intelligence, Surveillance, and Reconnaissance Center, known as C5ISR. Together, these organizations are conducting market research and technology trade studies to identify what the private sector can already deliver — and what may be ready soon.

At its core, the Army’s interest centers on a straightforward operational problem: soldiers tasked with breaching minefields and obstacle belts currently rely heavily on their own eyes and hard-won training. That process is slow, dangerous, and increasingly challenged by the sheer variety of explosive threats encountered on modern battlefields. Automated target recognition, or ATR, refers to technologies that augment or replace those human detection functions — using sensors and algorithms to spot, classify, and flag threats faster and with less risk to personnel.

The RFI does not specify a preferred sensor type, leaving the door open for both ground-based systems and unmanned aerial platforms. Companies offering both a sensor and an algorithm are eligible to participate in a live collection event, tentatively scheduled for the fourth quarter of fiscal year 2026 at Fort A.P. Hill, Virginia. There, systems will be evaluated against threat-representative explosive hazards in real field conditions, including both day and night collection periods. The Army has indicated it may provide a fixed amount of funding to participating vendors, though the specific figure has not yet been determined.

Companies that have developed detection algorithms but do not manufacture their own sensors are not excluded. A separate lab-based algorithm assessment is planned for later in 2026, allowing software-focused firms to have their systems evaluated against existing data sets within a government laboratory environment — at no cost to the Army.

The technical bar the Army is seeking to clear is significant. Any proposed ATR system must either match or exceed the probability of detection currently achieved by trained human operators, while simultaneously maintaining or reducing false alarm rates. In a combat breaching scenario, a false alarm can cost valuable time; a missed detection can cost lives. The Army is also requiring that proposed solutions either integrate with existing Army equipment or make a compelling case for why they should replace it entirely.

Explosive hazards have remained one of the most persistent and lethal threats facing U.S. ground forces across decades of conflict, from the roadside bombs of Iraq and Afghanistan to the dense minefields now documented across contested terrain in Eastern Europe. The scale and variety of those threats have only grown. Modern battlefields feature not just factory-produced anti-tank mines but improvised devices of enormous variety, buried or surface-laid across wide areas with deliberate intent to slow and attrit advancing forces.

Autonomous breaching — moving through obstacle belts with reduced human exposure — has become a priority across the Army’s modernization agenda. Combining robotic platforms with AI-driven detection could allow breach operations to proceed faster and with smaller exposed teams.