AI Edge Technology Demonstrated for Mine Detection in U.S. Army Exercise

Safe Pro Group Inc. has completed a U.S. Army live fire exercise in which its patented artificial intelligence was used to process drone imagery at the edge, enabling the identification of live scattered mines and supporting commanders with intelligence for mission planning.

The demonstration follows an initial U.S. Government contract for AI-enabled edge processing and the recent rollout of NODE-X, Safe Pro’s next-generation compact edge compute system for drone imagery analysis, threat detection, and manoeuvre support operations.

NODE-X is deployed as a backpack-based kit combining an edge compute unit with ruggedised GPU laptops running the company’s OnSight Windows software. The system is compatible with U.S. Army-approved Short-Range Reconnaissance (SRR) drones and is designed to perform onboard AI inference for threat detection and battlefield situational awareness.

Alongside detection capabilities, NODE-X supports the creation of 3D models, orthomosaics, and digital surface models. These outputs include terrain attributes such as vegetation height and slope analysis, enabling automated route planning and improved operational decision-making in the field.

Dan Erdberg, Chairman and CEO of Safe Pro Group Inc, commented, “We are honored to have been invited to work side-by-side with our Army stakeholders in the field during this high-pressure exercise. We believe that through these live minefield exercises, the Army is gaining increased confidence in our capabilities as we work together, gather soldier feedback, and deliver battle tested capabilities that our customers need fielded.”

“As we enter Q2, we are building strong momentum with our customers and are continuing to engage in discussions on how our AI technologies can help support their requirements in the short, medium and long term.”

Further U.S. Army exercises involving NODE-X are scheduled through the second quarter of 2026 as the company advances deployment and operational testing.

The Safe Pro Object Threat Detection (SPOTD) platform underpins the system, using machine learning models trained on a large-scale real-world dataset of drone imagery from Ukraine, consisting of 2.6 million images and 47,000 confirmed detections.

SPOTD is capable of recognising more than 150 categories of explosive threats, including landmines, cluster munitions, Unexploded Ordnance (UXO), and ambush drones. It converts raw drone video into high-resolution 2D and 3D geospatial models to support analysis and decision-making across defence, security, and humanitarian missions.

The platform is designed for edge operation without reliance on external connectivity, enabling rapid processing, mapping, and dissemination of geospatial intelligence in field environments.