Dick Bourgeois-Doyle

Ottawa-based writer and former Secretary General of the National Research Council of Canada.

Researchers develop drone-based wildfire early warning system

March 6, 2019 | 2 minute read

Images of crews scurrying beneath a dripping wall of flames and bombers tossing fistfuls into an inferno approaching homes and schools make it clear that the best way of battling a wildfire is, like fighting a war, not to have one in the first place.

This assumption underpins a new study reported in the Journal of Unmanned Vehicle Systems by Dr. Songsheng Li, a computer engineering researcher at Guangdong College of Business and Technology in Zhaoqing, China. Li is developing an autonomous early warning system using small drones that patrol forests, gather environmental data, and analyze the threat of fires before they erupt.

His work parallels other efforts to apply new technologies in monitoring the forest lands that cover almost a third of the planet. In places from Fort McMurray, Alberta, to Christchurch, New Zealand, wildfires are occurring more often, lasting longer, consuming more property, and threatening more lives. 

Researchers are refining advanced digital photography, robotics, and satellite imaging for more effective monitoring.  Yet all these techniques have limitations associated with cloud cover, pollution, distance, and time. Most only detect wildfires after they occur.

To predict and intervene before fires break out, precise data on environmental conditions such as humidity and temperature will need to be collected and analyzed in real time. But the vastness of the territories concerned and the need for uniformity make the option of using humans as the collection vehicles impracticable.

Li has thus tested a system using GPS information, unmanned aerial vehicles (UAVs), and a sophisticated programing tool known as Intelligent Flight Modes.

By combining the “Waypoint” and “Point of Interest” features of this tool, a drone or UAV can fly through a forest on a regular schedule along a route defined by the prior placement of sensors and data collection instrumentation.

The UAVs pick up data by communicating with the collection points through Bluetooth Low Energy (BLE) wireless systems and then return home to a base station that immediately uploads the data gathered from the forest. The data are reviewed and analyzed; if a threat is indicated it is conveyed to users in the form of a wildfire early warning message.

“A challenge lies in deployment, installation, and maintenance as each BLEDC (the sensor and collection device) has to be installed on a tree,” notes Li in describing issues faced in developing the technology. “I became interested in wildfire prediction when I started my PhD in Australia, but with limited resources I had to concentrate on theory, not applications, until after graduation.”

UAV systems also have limitations linked to distance, weather, and battery life, but Li’s system has demonstrated the possibility that UAV monitoring systems could be low-cost, general consumer products affordable to small communities and groups interested in protecting defined areas. Validation of the concept for broad application will require big data collection through the active involvement of national agencies and integration with deep learning and artificial intelligence systems.

Given the myriad of ways Mother Nature and human failings can initiate wildfires, there will always be a need for the identification of an active blaze by humans ready to fight the flames. But warning systems like those being explored by Li could save more property and people by helping to avoid the battle in the first place.

Read the paper: Wildfire early warning system based on wireless sensors and unmanned aerial vehicle in the Journal of Unmanned Vehicle Systems.

Dick Bourgeois-Doyle

Ottawa-based writer and former Secretary General of the National Research Council of Canada.