This particular project pairs an existing camera system with new technology. The use of a camera network to aid in wildfire response has gained traction in California, as well. Sonoma County also recently implemented an AI solution to mitigate wildfire risks.
The program involves collaboration between AFPD, the Pitkin County Sheriff’s Office, Pitkin County Telecommunication, Elk Mountain Technology and Pano AI.
This project was made possible through a private donation to the fire department from one Colorado resident, according to AFPD Deputy Chief of Operations Jake Andersen.
The donation allowed the agency to move forward with this to determine the benefit to taxpayers before making any substantial investment beyond employees’ time.
HOW THE TECHNOLOGY WORKS
The technology aims to detect smoke within minutes using high-definition cameras designed with AI in mind, according to Satyam.
Rafi Sands, who leads business development for Pano, stated that the company’s technology helps with two specific purposes: determining whether the smoke is from a fire and identifying the location.
Sands explained that the cameras rotate 360 degrees every minute, a feature that allows them to capture imagery at 10 different points. Edge computers located onsite stitch the imagery together in real time, allowing someone at a dispatch center the ability to see what is happening, look around, zoom in or play time lapses of the scene. This allows responders to “coordinate a more targeted and strategic response.”
Pano’s web software also offers a triangulation component — made possible through ample camera coverage — that identifies an incident's location with accuracy to save responders time, Sands added.
As Satyam described, when an incident is detected, a notification can be sent to the department through SMS or email. Operators can then easily pull up a visual of that and determine the next steps. Responders can confirm whether it is from a prescribed burn, monitor the fire’s progress and determine the best course of action for responding.
POTENTIAL FOR EXPANSION
The technology currently covers a range of territory from El Jebel to Independence, Andersen said.
While he describes the software as “fairly intuitive,” he also notes that there will be training opportunities as the project expands.
“There will definitely be training with our folks and whoever else wants access to this, because that's one of our goals with this is to be collaborative,” Andersen stated. “And it's going to cover more than just our fire district. And even within our fire district, there's law enforcement and other stakeholders that can really benefit from this.”
As Satyam elaborated, there is significant potential for scalability as incidents that occur in the monitored region will enable further training of the AI.
He added that as this industry — AI in wildfire detection — develops, there is a possibility for integrating multiple data sources. This could include satellites, weather data and more, enabling a more comprehensive approach.
“So then you can arm folks that are in emergency response with actionable intelligence, which can then [be used to] appropriately respond,” he said. “It also means a lot of safety, because people are able to look at the perimeter and understand how things are moving.”
INCREASING WILDFIRE RISK IN COLORADO
Last year’s climate accelerated Colorado’s need for innovative wildfire mitigation and early detection strategies after the East Troublesome Fire set a record for rate of growth in a wildfire in the state.
“In the year 2000, Mesa Verde National Park had a fire called the Bircher Fire, and that fire grew to 23,000 acres, and that was the largest timber fire in recorded Colorado history 21 years ago,” Andersen explained. “The East Troublesome Fire grew over five times that much in one day; to me that illustrates a total change of the game in just those two decades.”
