California firefighters are training AI to spot wildfires
Firefighters train a robot to scan the horizon for fires. It turns out that a lot of things look like smoke. From a report: For years, California firefighters have relied on an extensive network of more than 1,000 mountaintop cameras to detect wildfires. Operators stare at computer screens around the clock for wisps of smoke. This summer, as wildfire season begins, California’s premier firefighting agency is trying a new approach: training artificial intelligence software to do the work. The idea is to harness and deploy one of the state’s great strengths — expertise in artificial intelligence — to prevent small fires from becoming the kind that have killed dozens of residents and destroyed thousands of homes in California over the past decade.
Officials involved in the pilot program say they are pleased with the early results. In about 40% of cases, the AI software was able to alert firefighters to the presence of smoke before dispatch centers received the 911 calls. “It’s greatly improved response times,” said Philip Selig, chief of staff for the California Department of Forestry and Fire Protection, the state’s primary firefighting agency known as Cal Fire. In about two dozen cases, Selig said, the AI has identified fires for which the agency never received 911 calls. Fires were extinguished when they were still small and manageable.
After an exceptionally rainy winter, California’s fire season has not yet been as devastating as in previous years. Cal Fire has counted 4,792 wildfires so far this year, which is below the five-year average of 5,422 for this time of year. Perhaps most importantly, the number of acres burned this year was only a fifth of the five-year average of 812,068 acres. The AI pilot program, which began in late June and covered six Cal Fire command centers, will be rolled out to all 21 command centers starting in September. But the program’s apparent success comes with caveats. The system can only detect fires that are visible to the cameras. And at this point, humans are still needed to ensure that the AI software correctly identifies smoke. Engineers from the company that created the software, DigitalPath, based in Chico, California, monitor the system day and night, manually checking each incident that the AI identifies as a fire.