As wildfires become more frequent and severe, tech startups say they can help

Thanks to climate change, forest fires are getting worse. And as fires get bigger and more frequent, firefighters are stretched out.

Policymakers, activists and experts are pushing for a number of both long-term and immediate changes to try to alleviate this situation, from decarbonizing the economy to improving forest management in the world. risk areas. In recent years, tech startups have also entered the fray, claiming that emerging technologies like AI, machine learning, drones and sensors can help firefighters do more with less, or the same.

These startups, like Pano, which offers hardware and software solutions only for firefighters, or Edgybees, a geospatial intelligence company whose tool can help fight forest fires, have raised tens of millions of dollars to this day. Many are still in the early stages, leading pilots with various fire departments in the American West in an effort to both gain a foothold and refine their models for the community at large.

“We learned that in fact there had been calls from many areas, like the Gavin Newsom administration, from the firefighting community, from the utility industry, from academia, from non-profit organizations. better detect and respond to fires, ”Sonia Kastner founder and CEO told Emerging Tech Brew. “They were asking for cameras, drones, satellites and AI.”

For its part, Pano claims to have raised $ 6.5 million since its inception in 2020. The 20-person company is positioning camera networks in high-risk fire areas, which aim to identify and track fires. from forest to source.

Pano wants to place his cameras in high vantage points, be it mountains, radio towers, or better yet, radio towers on top of mountains. This would give a broad overview of forests at risk. Its cameras can then monitor those areas 24/7, transmitting video to an AI system trained to report minute-by-minute smoke plumes and fires. Human examiners are also in the know, confirming what the AI ​​is reporting and ultimately making the call to notify the fire departments.

The company’s ultimate goal is to deploy more than 5,000 cameras and detect forest fires before they get out of hand. But that’s just the beginning: Pano says it has deployed cameras to 22 locations in Oregon, California, Colorado, and Montana. And in Redwood City, one of the places where its technology is deployed, Pano mainly detected structural fires that the department was already aware of.

“The cameras can be used not only for detection with AI, but they can also be used for this essential confirmation step,” Kastner said. “We have a Pano Intelligence Center looking at all of these flows. They’re looking at our AI detections, they’re looking at satellite alerts, they’re looking at 911 feeds, and then they’re going to cross them with the camera feeds to locate the smoke.

Pano, who has conducted pilot projects with fire departments like the South Lake County Fire Protection District and the Big Sky, MT Fire Department, hopes his technology can streamline a confirmation process that , in its current state, is heavy and time consuming. result, dangerous.

Traditionally, fire detection has included the use of watchtowers or calls from passersby. When phone calls came in to report a forest fire, the fire departments would go to a high point in the area to visually confirm and locate the plume of smoke. This process is full of problems, especially the time-consuming nature of climbing a hill to spot a fire. Meanwhile, what was initially a small fire can turn into a big one, even before the fire departments know how to react.

“The fire department, we were going to one of the highest places we could… and we were driving up there and trying to figure out where the fire was,” Denise Enea, former Woodside Fire Marshal Fire Protection District in California, told Emerging Tech Brew. “Then we communicate with other teams trying to go to the fire and try to locate it that way. You can imagine how Neanderthal and how awkward it was, you know?

Other companies, like the forest fire-focused disaster management platform Cornea and geospatial intelligence firm Edgybees, are leveraging technologies such as drones and satellites to give firefighters a detailed look at developments. fires.

Edgybees, which has raised $ 16.4 million since 2017 and works with four fire departments in California, Florida and Australia, is focused on visually augmenting aerial and satellite imagery. It overlays the raw feeds with information such as street layouts, traffic lights and other key landmarks, which CEO Adam Kaplan says are “like Google Maps and Google Earth on steroids.” (For what it’s worth, Google Maps now also offers wildfire maps.)

These overlays, mixed with real-time drone footage, could allow firefighters to better understand the situation on the ground, allowing for a faster and more accurate response to fires, especially those that infiltrate residential areas.

Cornea was founded in 2018 and its fire chief is Tom Harbor, who has led the fire response for the US Department of Forest Services for over a decade. Cornea CEO Margot Preuss told us the company works with at least two “state-level fire departments in the west,” as well as the US Forest’s Rocky Mountain research station. Service and Colorado State University to refine the machine learning algorithms it uses to test and predict the size and extent of a wildfire as it grows.

One of Cornea’s two products involves aggregating information from places such as satellite imagery and historical fire data, and using it to then assign a fire extinguishing difficulty rating. Firefighters can then use the rating to make decisions about the allocation of resources and personnel. Cornea’s algorithm can produce these ratings in minutes, the company claims, saving time in a process where every second counts.

“This machine learning tool [is] Hopefully this will reduce the number of hours someone spends fighting a fire and then in the off season helps them control future fires with planning, ”Preuss said.

Effective technology could make all the difference when it comes to fighting fires, but providing an abundance of technology for firefighters is not a solution in itself; technology must be easy to use. Each company told us they worked directly with fire departments, training them in their technology so response times were quick and fires could be kept under 10-acre authorities like Cal Fire, the California Department. of Forestry and Fire Protection, which are essential. to control fires.

“The goal is to make the exit as easy as possible,” said Preuss. “They don’t need to know what is going on behind the machine learning algorithms, but they do need to know, as a first responder, where this fire is heading, what its scale and how it is. ” switch off as efficiently as possible. ? “