According to researchers at Montreal's Concordia University, the future of forest firefighting may lie in using artificial intelligence embedded in drones flying over high-risk areas. The idea is to be able to detect and geolocate forest fires as early as possible, before they get out of control.

Canada is regularly the victim of devastating forest fires, such as those that ravaged part of the country in 2023. In addition to reducing millions of hectares to ashes, they altered air quality in parts of Canada and the United States.

The key to controlling fires of this kind is to detect them as early as possible. The sooner they are detected, the easier it is to extinguish them.

Today, this mainly involves analysing satellite images and data provided by various ground sensors, or flying over entire areas by helicopter.

Researchers at Concordia University's Gina Cody School of Engineering and Computer Science have decided to develop a fully automated system incorporating artificial intelligence algorithms for processing satellite and/or drone images. Their aim is to be able to spot potential fire outbreaks and monitor their progress before they get out of control.

In this scenario, drones act as mobile sensors, deployed to analyze a specific perimeter. An initial analysis of satellite images identifies suspected smoke or flames, after which a drone is automatically directed to the site. Its cameras (visual and infrared) then analyse the progress of the fire. They can also estimate the distance to the fire, all of which is then communicated to the fire department so that they can take the best possible action.

In the future, it would be ideal for these same drones to be equipped with a fire-extinguishing system, although this appears to be very difficult, not least because of the weight of the water that would need to be carried.

The scientists' first field tests proved effective on a small scale, and their findings are published in IEEE Transactions on Industrial Electronics.

 - AFP