Fire and Rescue NSW uses AI drone to locate lost hikers

According to ABC, two men in their early 20s were reported missing at 7pm on Tuesday while hiking the Dead Horse Gap walking track in Kosciuszko National Park. ABC reports members from the Jindabyne Fire Station launched a remotely piloted drone equipped with thermal imaging and an AI detection system. ABC reports the drone detected the hikers about 500 metres off the walking track. Police and State Emergency Service volunteers contacted the hikers using the drone's built-in speaker and guided rescuers to their location; ABC reports the pair suffered mild exposure and declined medical treatment. ABC reports this was the first time the drone's AI detection system had been successfully used by Fire and Rescue NSW, and FRNSW Inspector and regional duty commander for the NSW Alpine area, Phillip Eberle, said, "It's definitely helped make what could have been a long-term incident into a very short-term incident."

AI-enabled thermal detection reduces the search space in wide-area, low-visibility environments by highlighting heat signatures for human review. For practitioners, integrating computer-vision classifiers with thermal sensors typically involves tradeoffs between detection sensitivity, false positives from rocks or animals, and battery/flight-time constraints for small drones. Voice-downlink or built-in speakers are pragmatic additions that convert detection into rescue contact, shortening response loops.

Observers following search-and-rescue deployments note an emerging pattern where field-proven AI assists responders by accelerating location and reducing manpower hours. Operational adoption depends on validated detection performance across terrain and weather, regulatory clearance for BVLOS (beyond-visual-line-of-sight) flights in some jurisdictions, and integration with existing emergency dispatch workflows.

Watch for FRNSW updates on system performance and any published metrics, follow regulatory changes affecting BVLOS drone operations in Australia, and look for broader adoption case studies from other emergency services documenting time-to-find and false-positive rates.