Forest Fire Detecting VTOL

The focus of this project was to build a UAV which can reliable and early detect forest fires and to test the concept on the island of Gran Canaria. For this purpose, a VTOL was developed, which is equipped with a thermal camera, flight control and radio links in order to automatically detect even smaller fires and transmit its GPS position to a ground station. Furthermore the software for reliable detection was developed.

The VTOL with thermal camera, flight control and radio links was set up in our workshop before the trip. The VTOL was dismantled and packed for the trip to Gran Canaria. On Gran Canaria we were able to arrange three flight days with the local authorities.

After arriving in Gran Canaria, the VTOL had to be reassembled and put into operation. During transport, various modules had come loose and no longer worked properly.

One of many checks before the first test flight was to check the center of gravity. The first test flight was then carried out. During the first flight, only vertical flight and not horizontal flight could be tested. The students from the Spanish vocational school El Rincon were there to learn important details for their own projects.

A successful vertical flight could be demonstrated on the second flight day.

We had another drone with us to test the sensors and for video and image recording. This should also serve as a backup. If the VTOL cannot fly, this drone should carry the sensors and demonstrate fire detection.

On the third flight day, the first demonstrations and presentations of the system took place in front of the Spanish students. All systems are initialized and the flight test is performed in front of an audience. It is launched vertically and flown horizontally. A fire detection was also successful by putting a little BBQ in the middle of the runway.

Spanish TV made a report based on our project and flight tests.

After successful test flights, there was a final discussion with the professors and the students.

Software

The software stack ran on the Raspberry Pi. To the Pi we connected a FLIR Lepton Module for thermal imaging. As for the algorithm we decided to go for a simple blob detection on the thermal image as the fire always has a big temperature and therefore color difference on a thermal image. The raspberry then asked the Pixhawk Flight Controller for its position and together with the position of the fire in the image calculated a GPS Position for the fire. This information was then sent to the ground station over a telemetry link and displayed on a map.