A green hexagon on a stick with a solar cell inside: This is what the sensors look like that the Berlin start-up Dryad can use to detect forest fires. 20,000 of these gas sensors are already in use worldwide, said Dryad boss Carsten Brinkschulte at the MWC in an interview with heise online.
Advertisement
Dryad's early forest fire detection system called “Silvanet” is already active at a total of 50 locations – primarily in southern Europe, but also occasionally in Germany. The sensor technology comes from Bosch.
The associated AI was trained in the laboratory using machine learning to detect gases produced by forest fires. This is intended to reliably detect carbon monoxide and hydrogen and avoid false alarms.
Supplement to image processes
There are various technical approaches for the early detection of forest fires. In California, for example, an AI system works with more than a thousand cameras to identify forest fires faster than humans. In several German federal states, the forest fire monitoring system “Fire Watch” is also trying to detect forest fires using cameras. The aim of such systems is to detect and contain emerging forest fires more quickly.
“All of these systems make sense,” says Dryad boss Brinkschulte. “Silvanet is a good complement to the other solutions.” Camera systems and satellite images can only detect smoke after a certain amount of time, but can cover large areas. The Dryad system detects smoke faster, in 1 to 60 minutes, according to Brinkschulte. Emergency services can react much more quickly than with other early warning systems.
For this to work, the Dryad sensors must be installed closely together. Brinkschulte speaks of one sensor per hectare. Together with the necessary infrastructure, this is quite expensive, which is why the Dryad system is particularly suitable for danger hotspots. Dryad is suitable for areas next to railway tracks and hiking trails where the risk of human-caused fires is particularly high.
Forest fire protection with mesh infrastructure
Dryad's solar-powered sensors use LoRaWAN for wireless data transmission. They communicate with multiple mesh gateways, which in turn communicate with border gateways. You transmit data to Dryad's Silvanet Cloud via 4G or, as an emergency option, 2G. There the data is evaluated by the AI and, in an emergency, automatically forwarded to the responsible authorities or municipalities.
Last year, forest fires in Europe were said to have destroyed 260,000 hectares of forest by mid-August, according to records from the European Forest Fire Information System (EFFIS). The Intergovernmental Panel on Climate Change (IPCC) assumes that the risk of forest fires will continue to increase as a result of advancing climate change.
Dryad's sensors have been in use since last year. Since then, there has only been one emergency, says Brinkschulte: there was a fire in a forest area in Lebanon. The Silvanet system sounded the alarm after 30 minutes.
“Internet of Trees” as protection against wood thieves
In the future, Dryad doesn't just want to detect forest fires. The mesh infrastructure set up in the forest areas can be combined with other sensors to open up further fields of application. The plan includes sensors that record soil moisture, measure growth or even recognize the noise of chainsaws in order to protect forest areas from wood theft, explains Brinkschulte. “We want to essentially build the Internet of Trees.”
(dahe)