Safe Pro wins $1.3m Army subcontract for AI landmine detection on UGVs
Safe Pro Group (Nasdaq: SPAI) has been awarded a $1.3 million subcontract from an undisclosed U.S. defence prime contractor to integrate its AI-powered threat-detection platform into U.S. Army autonomous Unmanned Ground Vehicles (UGVs). The Florida-based company said the work covers integration and testing of its computer vision and machine learning pipeline onboard the vehicles, targeting real-time detection of explosive hazards during on-the-move operations.
The contract is powered by Safe Pro's patented SPOTD (Safe Pro Object Threat Detection) technology, which the company says can identify more than 150 categories of explosive threat, including landmines, cluster munitions, unexploded ordnance and ambush drones. The underlying model was trained on a dataset the company describes as comprising more than 2.8 million drone images and over 50,000 confirmed detections gathered across more than 35,000 acres of terrain in Ukraine.
The technology
Safe Pro's AI pipeline was originally developed for drone-based aerial survey, where commercially available off-the-shelf drones feed imagery into the company's cloud-hosted inference engine on AWS. Porting that capability to UGVs represents a meaningful platform extension: ground vehicles operate at closer range and in more cluttered environments than aerial platforms, placing different demands on model latency and false-positive rates. The company did not disclose specific performance benchmarks, inference speed figures, or the UGV platform or programme it is supporting.
The prime contractor behind the award was not named in the release, which is common practice in U.S. defence subcontract disclosures, and Safe Pro did not confirm the programme office or contracting command involved.
Market context
The award places Safe Pro within a fast-moving segment of the defence technology market: autonomous ground vehicle integration for route clearance and counter-IED missions. The U.S. Army has run several UGV programmes in recent years, including the Robotic Combat Vehicle and short-range reconnaissance platforms, and route-clearance robotics have been a procurement priority since operational experience in Iraq and Afghanistan highlighted the human cost of manual mine-clearance.
On the AI-enabled detection side, Safe Pro competes with a broader landscape of computer-vision-for-defence startups and established defence primes that have invested heavily in autonomous sensing. The UGV market has attracted interest from companies including Shield AI, Ghost Robotics and several prime-contractor subsidiaries, though direct comparisons in the landmine-detection niche are harder to draw given the specialised nature of the sensor fusion and training-data requirements.
Safe Pro's use of Ukraine conflict imagery as training data is a material competitive differentiator, given how difficult it is to assemble confirmed, georeferenced detections at scale from active conflict zones. That dataset, if its provenance and labelling quality are as described, would be difficult for new entrants to replicate quickly.
Regulatory and commercial read-across
As a Nasdaq-listed company with U.S. Government contracts, Safe Pro operates under ITAR (International Traffic in Arms Regulations) restrictions that will govern how SPOTD technology can be licensed or exported to allied militaries. The company's boilerplate disclaimer notes that the U.S. Government retains reproduction rights under the project agreement, which is standard for federally funded R&D but limits Safe Pro's ability to commercialise derived intellectual property without restriction.
The $1.3 million subcontract is modest relative to Safe Pro's stated ambitions across defence, humanitarian and law enforcement markets, but it marks a concrete step from drone-based aerial survey toward ground-vehicle integration, broadening the platform's addressable mission set. Investors will watch for subsequent task orders, a named prime contractor, and whether the UGV integration reaches a fielded programme of record.