This post is also available in:
עברית (Hebrew)
Computer vision can help a drone recognize that it is looking at a vehicle, building or person, but recognition is not the same as identification. In security and defense operations, knowing that an object belongs to a certain category may not be enough. An autonomous system may also need to confirm exactly which vehicle it has found or whether a person belongs to a friendly force, including when GPS and communications are unavailable.
Israeli company Sodyo has developed RangeMark, a visual identification system designed to give drones that additional layer of certainty. It uses passive printed markers that can be detected by a camera and decoded into unique, machine-readable identities.
The system requires no transmitter inside the marker. It also does not depend on GPS, cellular connectivity, radio-frequency signals or a network connection to perform the identification. Instead, the drone searches its camera image and processes the visual pattern directly.
According to NextGenDefense, the technology is designed for detection at distances reaching up to one kilometer, although recent flight-test results demonstrated shorter ranges with specific marker sizes.
During those tests, a 100-centimeter marker was detected from up to 500 meters, while a 50-centimeter version was identified at distances reaching 250 meters. The system searched for markers across a camera field of view covering approximately 20 hectares, rather than requiring the aircraft to point directly at a known location.
Testing included hovering, normal transit and search-pattern flights at speeds reaching 72 kilometers per hour, demonstrating identification while the drone was moving rather than only from a stationary position.
The system supports two marker formats. Color Code uses a high-density colored pattern intended for visible-light environments, while Carbon Code uses a two-tone design for more challenging visibility conditions.
For defense applications, the distinction between recognition and identification can be significant. AI might determine that an image contains a military vehicle, for example, while a machine-readable marker could establish which particular marked vehicle it is. Similar markers could potentially help identify friendly personnel, designated locations or equipment without requiring those assets to broadcast an electronic signal.
That passive approach could also be useful in environments affected by communications disruption or electronic warfare. Because the marker itself emits nothing, it does not add a radio-frequency transmission that could potentially be detected.
The system was showcased at Commercial UAV Expo 2026 and remains dependent on the target carrying a compatible marker. It therefore does not replace conventional computer vision for identifying unknown objects.
Instead, the technology addresses a narrower problem: when an autonomous machine sees something that has been deliberately marked, give it a way to move from “I know what kind of object this is” to “I know exactly which object this is.”
