Summertime is forest fire season – detecting fires with a flying camera swarm
Focus: Swarm exploration, disaster relief
Germany is sweltering. With national average temperatures well above 30 degrees Celsius, summer 2019 has really been taking a toll over the last few weeks. Over long periods, heat and dry weather invariably increase the risk of forest fires. The exceptionally hot summer of 2018 saw devastating fires in Greece and California that broke all records. In flight experiments with a swarm of drones, which were conducted as part of the HEIMDALL project, the German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt; DLR) demonstrated that aircraft equipped with cameras and thermal imaging sensors can detect small forest fire hotspots. The drones are able to explore a region autonomously and exchange data with one another.
In March 2019, the DLR Institute of Communications and Navigation carried out some initial experiments at three different locations in Catalonia, Spain. The aim of the first experiment was to detect a controlled fire that was being monitored by the Catalonian fire brigade. Such fires can prevent widespread forest fires. The swarm consisted of three autonomous Unmanned Aerial Systems (UAS) and gathered georeferenced data using visible-light cameras and thermal imaging cameras.
Further experiments demonstrated the detection of forest fire hotspots with a diameter of between 15 centimetres and 1.5 metres. Wood, straw and sepiolite were used for the hotspots. The cameras recorded the various phases of the hotspot dynamics – from ignition to the extinguished fire. In total, 37 different hotspots were successfully analysed during 19 flights at altitudes of between 30 and 130 metres.
The experiments showed that an autonomously operating swarm of drones can be used to detect forest fires. The UAS were able to identify even very small hotspots of less than 15 centimetres across from an altitude of 90 metres. The collected data will enable the improvement of the DLR’s algorithms for the reliable detection of forest fire hotspots by swarms of drones. The final demonstration of the DLR drone swarm is planned for March 2020, again in Spain.
Swarm exploration at the DLR Institute of Communications and Navigation
Experts at the DLR Institute of Communications and Navigation have been working intensively on the development and testing of exploration algorithms for autonomous robot swarms since 2014. The aim is to enable multiple robotic platforms to cooperate in order to produce a map of their environment without human interaction. To this end, the studies mainly focus on the challenges of developing new algorithms and methods for swarm exploration and the testing of these algorithms under realistic conditions using swarms of robots.
About the HEIMDALL project
The HEIMDALL project, which is funded by the European Commission as part of the Horizon 2020 programme, aims to develop and test new techniques for addressing complex, hazardous scenarios. One of the aspects considered within HEIMDALL is the detection of forest fire hotspots using a swarm of drones. To this end, a flexible platform for the management of complex hazard situations will be developed during the HEIMDALL project, offering innovative technology for response plans and emergency management support.