LONDON (AP) — Wildfires caused by climate change are Destroyed communities on Maui to mediterranean Many people have died this summer, firefighters are exhausted and new solutions are in high demand. Enter artificial intelligence.
Firefighters and startups are using AI-enabled cameras to scan the horizon for signs of smoke. A German company is building a constellation of satellites. Detecting fires from space.And Microsoft Use AI models This is to predict where the next fire will occur.
and Wildfires are getting bigger and more intense As the world warms, firefighters, utilities and governments are scrambling to use fire to stay ahead of fires. Latest AI technology — it caused both fear and excitement life changing possibilities. First responders are increasingly nervous and want AI to give them an edge, but they still need humans to make sure the technology is accurate.
Major firefighting agencies in California This summer, we began testing an AI system that looks for smoke in footage from more than 1,000 mountaintop cameras, and we’re now expanding it across the state.
The system is designed to spot “anomalies” and alert an emergency command center, where officials can check to see if it’s really smoke or something floating in the air.
“The benefit of this is that it immediately appears on the screen and the dispatcher or call taker can interrogate that screen to determine whether or not to send a crew,” said California Forestry Department Chief of Staff for Information. Philippe Selegh said. Fire protection.
Cameras are part of the network that previously required employees to monitor and provide billions of bytes of data. Data for AI systems To digest. Although humans still need to confirm smoke sightings, the system reduces fatigue for staff who typically monitor multiple screens and cameras, ensuring they only see them when there is a possibility of fire or smoke. Selig said it can help warn people.
It’s already useful. The battalion chief received a smoke alarm in the middle of the night, confirmed it on his cell phone, and called the command center in San Diego to dispatch first responders to the remote location.
Dispatchers said that without the alarm, they likely wouldn’t have known about it until the next morning, and the fire would have been even bigger, Selig said.
Pano AI, a San Francisco startup, is taking a similar approach, installing cameras on cellphone towers to scan smoke and alert customers such as fire departments, power companies and ski resorts.
The camera uses computer vision machine learning, A type of AI.
“They’re specifically trained to detect smoke, and they train with smoke images and non-smoke images,” CEO Sonia Kastner said. .
These images are combined with other data sources, such as feeds from government weather satellites that scan for hotspots, and social media posts.
This technology avoids one of the major problems with traditional methods of detecting wildfires. That means relying on 911 calls from passersby, which must be verified by officials before workers or water cannon planes are dispatched.
“Typically, only one in 20 of those 911 calls actually indicates a wildfire. Even during fire season, it could be clouds, fog, or even a barbecue. ”Kestner said.
Pano AI’s system still relies on a final check, where administrators play back a time-lapse of the camera feed to confirm smoke is rising.
“Technology is really becoming essential” to combating wildfires, said the executive vice president of energy distribution at Portland General Electric, Oregon’s largest utility and a Pano AI customer. President Larry Beckedahl said.
Utility companies sometimes play a role in starting wildfires. power lines are down Blown by the wind or hit by a fallen tree. Hawaii’s power company Power lines spark devastating fire on Maui It seems that it was knocked down by strong winds this summer.
PGE, which supplies power to 51 cities in Oregon, has deployed 26 Pano AI cameras, which Beckedahl said help speed response and coordinate with emergency services. said.
Previously, the fire department was “running around looking for something, but we didn’t even know exactly where it was,” he said. The cameras help detect fires faster and get teams to the scene faster, reducing response times by up to two hours.
“This is important because of how quickly fires can spread and spread,” Beckedahl said.
Using AI to detect fire smoke is “relatively simple,” said Juan Lavista Ferrez, chief data scientist at Microsoft.
“What’s not easy is having enough cameras to cover enough areas,” he said, referring to the vast, remote areas of northern Canada that were hit by fires this summer.
Ferres’ team is Microsoft is developing AI models Predict where fires are likely to occur. They fed maps of previously burned areas into the model, along with climate and geospatial data.
This system has limitations and cannot predict random events like lightning strikes. However, you can scrutinize historical weather and climate data to identify patterns, such as areas that are generally dry. Ferres said even roads with people nearby are a risk factor.
“Not everything is going to be perfect,” he says. “But what it can do is build probability maps based on what happened in the past.”
The technology, which Microsoft plans to offer as an open source tool, could help first responders trying to figure out where to focus their limited resources, Ferres said.
Another company is looking to heaven for a solution. German startup OroraTech analyzes satellite images with artificial intelligence.
Using advances in camera, satellite, and AI technology, OroraTech launched two small satellites, each about the size of a shoebox, into low-Earth orbit about 550 kilometers (340 miles) above the Earth’s surface. The Munich-based company has ambitions to send eight more spacecraft next year and eventually 100 into space.
as Wildfires hit central Chile Earlier this year, Ororatec announced it was providing thermal images at night, when drones are less frequently used.
Weeks after OroraTech launched its second satellite, it detected a fire near the Keg River community in northern Alberta. Remote stretching with fire This summer, I repeatedly observed the boreal forest.
CEO Thomas Grubler said: “The satellite is equipped with very efficient algorithms that allow us to detect fires much earlier.”
The AI also takes into account vegetation and humidity levels to identify flare-ups that can cause catastrophic large-scale fires. The technology could help stretched fire agencies direct resources to fires that can cause the most damage.
“Because we know exactly where the fire starts, we can see how it will spread,” Grubler said. “So which fires become big fires within a day, and which ones go out on their own?”
___
AP Technology Writer Barbara Ortutay in San Francisco contributed.