South Africa’s science research hub, the Council for Scientific and Industrial Research (CSIR) has partnered with a local drone manufacturing start-up to use drone technology in detecting all kinds of fires and even putting out small ones.

This, after South Africa has experienced recent several devastating and costly wildfires in recent years that the CSIR reckons would have been prevented through early detection and combatting.  

But since the organisation is currently not involved with any drone development, they roped in the help of Cape Town-based drone manufacturer, Autonosky, who were already working on a drone for fire-fighting operations.

Aside from detecting fires, Autonosky have a made drone that is capable of dropping fire-retardant balls to douse small flames.

The CSIR then added fire-detecting sensors to the drone.

“It is unique in the sense that we are not detecting fires in the normal sense,” Edwin Magidimisha, a senior physicist at the CSIR told broadcaster SABC recently, in explaining how the sensor works.

“We are using the near infra-red band giving us a vast list of available high-resolution CCD/ CMOS (charge-coupled device and complementary metal oxide semi-conductor) detectors compact enough to fit on a nanosatellite or small-sized drone. CCD and CMOS are two different types of imaging sensors.”

“Traditional methods for the detection of fires have been through the use of the infra-red band, using heat.  Choosing to use the near-infra-red band was because plants when they are growing, absorb natural elements from the ground. When those plants burn during the flaming phase of the fire, they emit energies at specific frequencies, which we harness to detect the presence of fire.”

On how a typical operation would work, Magidimisha said; “If we have been asked by a particular province, to pick up a fire, we fly the drone to a specific location to search for fires or where a fire has been reported.

“If we pick up the fire, we will have to send notification to the ground station that the fire has been detected at these GPS co-ordinates, with images of the fire.  If the drone is hovering over the fire, we also provide information in which direction the fire is progressing based on the wind direction and fire size.

“This information should not only be provided to the ground station but also to cell phones to provide alerts for the firefighters. While the firefighters are combatting the blaze, the drone should also continuously monitor the fire and provide real-time information on any changes. The drone will also be able to drop fire-retardant balls to prevent a small fire from expanding.”

He added that the camera and payload prototype was initially developed around 2017 and was launched with the ZACUBE 2 satellite as a secondary payload in 2018. This was to test the satellite application system’s performance from space; and the lessons learned included detecting and eliminating false alarms. 

These results led the CSIR team to believe that the concept would also work well on a terrestrial application like a drone.

As Magidimisha explained; “For example, when the satellite makes an overpass over a region closer to water bodies, how can one deal with that; because reflections from water bodies can impact the sensor?

“We are using ultra-narrow band filters centred at a specific band together with the band differencing technique to get rid of information that is not necessarily related to fire detection.

“So far the results show that we are actually detecting fire much better.”

The drone has been tested at the Centurion Flying Club, in Tshwane, flying it at short distances to over 200m altitude. “We were able to detect fires over those ranges. The range limit was because of the live streaming telemetry (the collection and analysis of data while the system is airborne) from the payload to the ground. 

“But the results over 200m were also encouraging.  We have also run other tests in the Cape where we were able to detect small-scale fires, some in between trees, and some in an open space.  We were able to detect them with no noise at all.

“We are now trying to see if we can integrate on-board image processing so we can only send fire alerts to the fire-fighters without any post-image processing required.”

However, he highlighted that the challenge now is to find interested partners who can make full use of the technology; adding that the CSIR is currently in talks with the national parks in the Cape region.

Autonosky are the makers of the Autono 1, a multi-purpose drone that cane be used for applications that include package delivery, search and rescue, surveillance and other data collection missions.

Drones have been the technology of choice in fire-fighting efforts in California, USA, where forest fires rage in summer. The drones have been used to detect and monitor the fires, to give firefighters and accurate reading of what they are up against.