{"id":22758,"date":"2026-10-12T17:00:31","date_gmt":"2026-10-12T21:00:31","guid":{"rendered":"https:\/\/www.autonocion.com\/us\/?p=22758"},"modified":"2026-10-12T17:04:18","modified_gmt":"2026-10-12T21:04:18","slug":"386-pound-german-fire-drone-alaska","status":"publish","type":"post","link":"https:\/\/www.autonocion.com\/us\/386-pound-german-fire-drone-alaska\/","title":{"rendered":"A 386-pound German drone carrying 26 gallons through a standard fire hose nozzle flew itself to a fire in Alaska after 170 solar-powered sensors in the trees smelled it, landed its load right on the flames and took second place behind an American defense company"},"content":{"rendered":"

Back in July, smoke from wildfires in Canada had people in New York, Boston, and even Washington, DC, being warned about the air they were breathing. During that time, we were busy telling you to go check the cheap filter behind your glovebox<\/a>.<\/span><\/p>\n

But a cabin filter only deals with smoke that’s already in the air. A German company called Dryad Networks wants to catch fires while they’re still small enough for a drone to handle. And in June, it tried that out in the Alaska wilderness with a 386-pound drone nobody was flying.<\/p>\n

It finished second. XPRIZE named the winners<\/a> of the autonomous track in its $11 million Wildfire competition on September 23 at the New York City Fire Museum (a pretty on-the-nose venue). On October 8, the company laid out what its system actually did in Alaska.<\/p>\n

Anduril Industries, the American defense company, took first place and $1.2 million, plus a separate $1 million bonus from Lockheed Martin for how quickly and precisely it spotted fires. Dryad got $800,000. A team called AURA Foresight took $500,000 for third.<\/p>\n

So what did XPRIZE ask for?<\/h2>\n

XPRIZE didn’t make this contest easy.<\/p>\n

In June, the finalists set up near Nenana, a small town an hour or so southwest of Fairbanks. Each system had eight hours to watch over 386 square miles of Alaska backcountry, with up to three controlled fires to handle. It had to find a high-risk fire and put it out completely within 10 minutes of ignition, and it wasn’t allowed to touch the decoy fires. The teams couldn’t step in at any point. XPRIZE compared the area to San Francisco, Oakland, and San Jose put together.<\/p>\n

Decoy fires are pretty much what they sound like. They’re fires the system’s supposed to ignore. In the real world, you’d be looking at a campfire or a burn pile somebody’s keeping an eye on, and a drone that hoses down every one of those wouldn’t stay popular for long.<\/p>\n

Plenty of drones already handle dangerous jobs, like the caged drone that mapped radioactive waste vaults in Idaho<\/a>. But people were at the controls for that one. In Alaska, each system had to decide by itself when to launch, where to go, and when to let go of its load.<\/p>\n

It takes two drones and 170 sensors<\/h2>\n

Dryad’s system starts on the ground, with sensors it calls Silvanet. Each one goes up on a tree or a pole, runs on a small solar panel and supercapacitors, and talks to its neighbors over a wireless mesh network. XPRIZE says the company had 170 of them out in Alaska.<\/p>\n

Each sensor is basically a smoke detector for the woods, much like the one on your ceiling, but with a built-in gas sensor. Wood starts giving off gases as soon as it gets hot enough, before it actually bursts into flames, and the sensors sniff the air for the carbon monoxide, other gases, and fine particles that come off a fire at that smoldering stage.<\/p>\n

There’s no lithium battery inside, either, and I’d call that a pretty smart move for a gadget that spends its life in a forest that might burn.<\/p>\n

The drone side of the system is called Silvaguard. When a sensor detects something, a small observation drone takes off on its own. The company lists it at about 30 pounds at takeoff, with up to an hour in the air and a top speed around 65 mph, and it carries thermal and regular cameras plus lidar, which is basically a laser scanner. It finds the fire, determines whether it’s real or a decoy, and pinpoints exactly where it’s burning.<\/p>\n

Then there’s the suppression drone. It’s a commercial heavy-lift quadcopter that Dryad fits with its own tank, a high-throughput pump, and, of all things, a standard fire hose nozzle. The company’s spec sheet<\/a> lists a maximum takeoff weight of 386 pounds and a tank capacity of 26 gallons. However, the spec sheet doesn’t say whose airframe the company started with, or how big the finished drone is.<\/p>\n

\n
\n
Suppression drone<\/div>\n
386 lb<\/div>\n
Maximum takeoff weight of the heavy-lift quadcopter, per Dryad’s spec sheet.<\/div>\n<\/div>\n
\n
Suppressant tank<\/div>\n
26 gal<\/div>\n
Fed by a high-throughput pump to a standard fire hose nozzle.<\/div>\n<\/div>\n
\n
Ground sensors<\/div>\n
170<\/div>\n
Solar-powered gas sensors in Dryad’s network at the Alaska finals, per XPRIZE.<\/div>\n<\/div>\n
\n
TARGET<\/div>\n
XPRIZE goal<\/div>\n
10 min<\/div>\n
From ignition to a fire that’s completely out, across 386 square miles of Alaska, with no human help.<\/div>\n<\/div>\n
Figures: Dryad Networks, XPRIZE.<\/div>\n<\/div>\n

That 386 pounds is the heaviest the drone’s designed to be when it leaves the ground, payload included. Water weighs about 8.3 pounds per gallon, so if the tank’s full of water, that’s roughly 220 of those 386 pounds. The company doesn’t spell out its suppressant mix, though XPRIZE’s photos show the drone dropping water during a live demo for guests at the finals.<\/p>\n

So why not just give it a bigger tank?<\/p>\n

Because every<\/em> extra pound makes a drone with rotors work harder just to stay in the air, and that eats into its battery. The tank isn’t that big to begin with, either.<\/p>\n

Its 26 gallons come to about half of what the 50-gallon water heater you might have in your basement holds, and Dryad CEO Carsten Brinkschulte is fairly upfront about what that means. He says drones can knock down a small ignition but don’t carry enough to matter once a wildfire gets big, so the whole system’s built around catching fires while they’re tiny.<\/p>\n

So did anybody pull off the whole thing?<\/h2>\n

Not as far as I can tell. XPRIZE had said a team that won a track outright could get $3.5 million, and none of the three finalists took that home. The judges split $2.5 million among them instead, as credit for progress on different parts of the job.<\/p>\n

XPRIZE singles out third-place AURA Foresight for clearing a 10-minute cutoff for detection, but its write-up doesn’t say any team got a fire all the way out in time. Adam Broecker, a Lockheed Martin vice president, said in PG&E’s announcement of the awards<\/a> that “the final prize remains on the horizon.”<\/p>\n

Dryad’s being careful about how it describes its own run, too. The company says its sensor network found and confirmed the fire within 12 minutes, set off the response on its own, ignored the decoys, and that the suppression drone landed its load right on the fire.<\/p>\n

Dryad calls that a proof of concept and says the results show where more work’s still needed. The company hasn’t broken that 12 minutes down, though, so I can’t tell you how long it took from ignition to the moment the suppression drone hit the fire.<\/p>\n

Honestly, I don’t think any of that’s a knock on the teams. XPRIZE asked machines to find a fire and put it out completely within 10 minutes, with no one touching anything, and billed the whole thing as a challenge no one had taken on at this scale.<\/p>\n

The three teams went at the problem in very different ways. XPRIZE had named five finalists in January, and only three competed in Alaska, one of them a late addition. Anduril tied everything together with Lattice Fire, its own software. A Sentry tower watched from the ground, third-party VTOL aircraft (they take off and land vertically) scanned from above, and the closest Ghost-X drone got sent out to drop fire-retardant balls on the fire, according to XPRIZE’s coverage of the finals<\/a>.<\/p>\n

AURA Foresight, which combined a British semifinalist team with an Australian one, used cameras on boom lifts and commercial DJI drones. XPRIZE quotes co-lead Rob Vernon calling it the “Home Depot approach.” Dryad, meanwhile, showed up with a flying fire hose.<\/p>\n

Real forests come next<\/h2>\n

The company says it’ll now invest serious money to turn Silvaguard into an actual product. XPRIZE is moving into what it calls an Impact Phase, built around field trials, and PG&E, the California utility, is the lead sponsor. PG&E says that phase will run 12 to 15 months.<\/p>\n

You won’t see one of these over an American forest without some paperwork, though. The FAA rules most commercial drone pilots fly under don’t cover drones that weigh 55 pounds or more, so a 386-pound drone flying itself over a forest needs special approval. The Alaska finals were based at Nenana’s municipal airport, and the University of Alaska Fairbanks drone center, one of the FAA’s designated test sites, handled the airspace.<\/p>\n

Frankly, I wouldn’t be shocked if the paperwork took longer than the engineering.<\/p>\n

“Wildfires are a race against time,” Brinkschulte said in the October 8 announcement. Dryad says it’s planning real-world pilots for 2027, and it wants the drones on the market in 2028.<\/p>\n","protected":false},"excerpt":{"rendered":"

Back in July, smoke from wildfires in Canada had people in New York, Boston, and even Washington, DC, being warned … <\/p>\n

Read more<\/a><\/p>\n","protected":false},"author":9,"featured_media":22763,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[121],"tags":[],"class_list":["post-22758","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-industry","resize-featured-image"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts\/22758","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/users\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/comments?post=22758"}],"version-history":[{"count":5,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts\/22758\/revisions"}],"predecessor-version":[{"id":22815,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts\/22758\/revisions\/22815"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/media\/22763"}],"wp:attachment":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/media?parent=22758"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/categories?post=22758"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/tags?post=22758"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}