A solar sensor mesh caught Alaskan test fires before open flames appeared

In a 1,000-square-kilometer stretch of Alaskan wilderness this June, a 170-sensor solar mesh from Berlin-based Dryad Networks caught deliberately set fires during the smoldering phase, before open flames appeared, by sniffing their gas signatures. Two autonomous drones then put them out within 12 minutes, without a human in the decision loop, earning the $800,000 top prize at XPRIZE Wildfire Track B.

The Silvanet sensors act as an electronic nose: solar-powered gas detectors that flag wildfire signatures during smoldering, the stage where heat is rising but flames have not broken through. Small ignitions are still suppressible at that point, so a drone with retardant can actually finish the job, something no autonomous system can do once a fire has reached crown stage. Sensors form an off-grid mesh that does not depend on cellular coverage.

The June finals required finalists to detect and fully suppress high-risk fires within 10 minutes of ignition across the 1,000 sq km test range, while leaving planted decoy fires untouched. Dryad's Silvanet mesh, its Silvaguard observation drone, and its suppression drone cleared that bar. Track B awarded $2.5 million total across three finalists: Dryad, Fire Foresight, and AURA Foresight.

Deployment is already underway: Dryad has manufactured more than 40,000 sensors, with recent deliveries of 10,000 to France and 7,000 to Italy.

CEO Carsten Brinkschulte framed the win as a proof point for the smoldering-phase approach, though the Alaska test was a set-piece with deliberate targets. Independent field reliability, false-alarm rates, and decoy rejection under real weather remain unverified.