Researchers at the University of São Paulo’s São Carlos Engineering School (EESC-USP) are developing drones equipped with gas sensors and artificial intelligence to help detect forest fires.

Fire prevention and control agencies in São Carlos, in the state of São Paulo, Brazil, will soon be able to use these drones to identify and respond to fires early, before they spread.

The project was presented last week, during the aeronautics session of Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, which translates to Sao Paulo Research Foundation) Week Toulouse, which took place from June 10 to 12 in Toulouse, the capital of the Occitanie region in southern France.

We’re already working with the Civil Defense, the city government, and the São Carlos Department of the Environment, and we’ve submitted a proposal to evaluate using the drones we’re developing to identify fire hotspots in the municipality,” said Glauco Augusto de Paula Caurin, Professor and Project Coordinator at FAPESP

The drones are equipped with small, low-cost sensors developed by the research team. These sensors can selectively detect and continuously monitor levels of carbon dioxide and methane in the air flowing through the drone. They also measure other environmental factors such as temperature and humidity.

“We made several adjustments to integrate specific gas sensors that work together like an electronic nose,” de Paula Caurin added.

The sensors collect data on gas concentrations, which are then processed by artificial intelligence algorithms to identify emission sources. This allows the drones to detect the presence of carbon dioxide and trace gases like methane released during fires as they fly over an area.

“Drones allow forest fires to be detected much more quickly than satellites can, for example. This enables the authorities to respond faster and control them more quickly.”

A recent study supported by FAPESP through the Research Centre for Greenhouse Gas Innovation (RCGI), in partnership with oil company Shell, evaluated the use of drones for monitoring greenhouse gas (GHG) emissions.

According to the researchers, the results show that drones are more efficient and cost-effective than existing tools such as satellites, research aircraft, and observation towers.

de Paula Caurin said they compared data collection from a single aircraft flight to multiple drone flyovers. Drones also offer better control over targeting specific areas for data collection compared to satellites, which may only revisit the same location every two days.

Another advantage of using drones is the ability to adjust the altitude of data collection.

“By collecting greenhouse gas data with drones, instead of an average of carbon dioxide or methane on a surface, it’s possible to obtain the volume of distribution of these gases in a given region,” said de Paula Caurin.

“Today, even with the best satellites, we’re unable to obtain this volumetric information. With drones, greenhouse gas data collection is no longer a surface or map of the region, but rather volumetric information.”

The study notes that current commercial drones can fly for only 15 to 30 minutes, which limitation makes them unsuitable yet for covering large areas such as forests.

To address this, the research group plans to improve drone efficiency through aerodynamics studies. The goal is to extend flight range and increase coverage.

The team tested drone-based GHG monitoring at the University of São Paulo campus in São Carlos. The location sits in a transition zone between the Atlantic Rainforest and the Cerrado, a savanna-like biome in Brazil.

“In the future, with more suitable equipment, we intend to carry out missions in the Amazon,” de Paula Caurin concluded.