Ukraine has tested the Safe Pro artificial intelligence system built into Black Widow drones for the first time. The technology is capable of recognizing mines, cluster munitions, and other explosive objects in real time.
Ukraine has tested the Safe Pro artificial intelligence system built into Black Widow drones for the first time. The technology is capable of recognizing mines, cluster munitions, and other explosive objects in real time.
Militaryi reports on the cooperation and testing. It involves the integration of the SPOTD (Safe Pro Object Threat Detection) complex, which processes 4K video on board the Black Widow and identifies more than 150 types of threats in real time: from anti-personnel mines and cluster munitions to unexploded ordnance. The detected objects are automatically transmitted to military situational awareness platforms, in particular the US Army’s ATAK ecosystem, which speeds up tactical decision-making.
A separate module, SPOTD NODE (Navigation, Observation and Detection Engine), creates interactive 2D/3D threat maps and allows you to work on the front line and in environments with unstable communication. Data is synchronized between units, so sappers, assault teams and logisticians receive a single «big picture» of risks.
The uniqueness of the approach lies in the real Ukrainian data on which the model was trained. According to Safe Pro, the dataset included 1.88 million UAV images covering 8,119 hectares of territory; the algorithm recorded more than 34,200 explosive objects. This is one of the largest bases in the world for mine action and post-combat clearance tasks.
«The battlefield is changing rapidly, and the military needs enhanced real-time situational awareness. Combining Black Widow with Safe Pro technology delivers just that,» said Dan Erdberg, CEO of Safe Pro Group. Red Cat added that they are aiming for rapid deployment of the solution for the US Army and allies, as well as compatibility with existing unit workflows.
For Ukraine, one of the most mined countries in the world, such systems are not only about tactics on the front line, but also about accelerating humanitarian demining, safer evacuation routes and restoring critical infrastructure. The integration of detection on board UAVs reduces the time between «seen» and «marked on the map», and a single threat map allows commanders to plan operations and logistics with fewer losses. If serial deliveries go to the troops, the next step will be to scale up the training of algorithms on new regions and types of threats — from booby traps to improvised explosive devices.
Previously, dev.ua wrote about how over 50 Ukrainian IT specialists in Kyiv joined forces at the AI Data Jam hackathon to train artificial intelligence to recognize explosives in aerial photographs. The initiative was a continuation of the spring AI Labeling Sprint and aims to accelerate humanitarian demining of Ukraine’s territories.
