Extinguishing Wildfires in Large Scale Scenarios Using Swarms of UAVs

The climate crisis induces the appearance of wildfires. Identifying and mitigating them at an early stage is crucial to control them successfully. To achieve this, we present algorithms to mitigate different types of wildfires using swarms of high-payload UAVs. In our experiments, a swarm of 30 UAVs monitors and suppresses wildfires in an area as large as California using a newly developed algorithm called Dynamic Space Partition for Firefighting (DSPF). We test the algorithm in two different environmental scenarios from low to high-difficulty fire conditions. We created DSPF with coordination (DSPFC) to enable multiple UAVs to engage larger fires. Using this algorithm, the aircraft that identified the firefront communicates with the two closest aircraft to self-organise and engage the wildfire. We developed a metric named ‘fire mitigation effectiveness’ (FME) to compare the different algorithms. Our results show that the DSPF was able to mitigate an average of 82% of the wildfires and achieve an FME of 61% at a low-difficulty scenario. When facing a high-difficulty scenario the DSPF strategy mitigated an average of 18% of fires, achieving an FME of 18%. The DSPFC achieved a better performance compared to the DSPF in the high-difficulty scenario mitigating an average of 73% of fires and achieving an FME of 50%.