Drones being developed to track bushfires

Drones are cheap. They’re plentiful. They’re flexible. And the real-time surveillance information they deliver can be a dramatic “force multiplier” for an embattled ground commander needing to know where to best deploy their limited force to fight a bushfire.

And that’s just in the military sense.

The Australian National University is exploring the potential for the technology – be it a line-of-sight quadcopter or over-the-horizon fixed-wing craft – to become as much a part of the nation’s firefighting arsenal as a high-pressure hose and CB radio.

“We had an open day just a couple of weeks ago where we demonstrated our Carbonix vehicle to give them an idea of how it works,” says Associate Professor Andrew Tridgell.

A local ACT fire brigade set up a campfire on the ANU’s Spring Valley Farm experimental robotics facility near Canberra. And while the autonomous-capable drone remained close to its controller, it proved its ability to sense, locate and identify a fire source.

“It was just a quick demonstration, but there were quite a few people from the fire services who turned up. And there was some very positive feedback,” the ANU computer scientist told Cosmos.

“We need to change the way we fight bushfires,” Tridgell says.

Small drones with thermal cameras are already a common firefighting tool. “I’ve spoken to plenty of people – forestry services and local fire brigades – that have a drone available.

“They’re incredibly cheap for what they can do. They can fly for 30 or 40 minutes and provide a real-time thermal video feed … it’s like a flying fire tower.”

But, by that stage, firefighters are usually already battling a fully involved blaze.

The drones can help, but they’re not a complete solution.

“It’s about being able to look around and see a new spot fire developing over there, and if the other one is dying down,” Tridgell explains. “I’ve got a certain number of appliances and firefighters. I’ve got a certain amount of water. I’ve got aerial support I can call in – crewed helicopters or tankers. So, for professional and volunteer firefighters alike, having that eye in the sky gives me better situational awareness when it comes to using what we have.”

But spotting an early outbreak is a whole different matter.

“Once a fire gets really large, the only way it really gets put out is rain,” says Tridgell. “So one of the key aims of this project is to get to the fires when they are very, very small.”

Australia is, however, a big place. A single dry lightning storm can generate thousands of strikes over hundreds of square kilometres. And several such storms can happen in the space of a week.

“In some areas, the most common way that fires are detected is through a member of the public noticing smoke and calling 000,” he says. “And commonly, that triple zero call might come 12 to 18 hours after the fire had started. Really, we should be able to do better than that.”

Currently, using a light aircraft or helicopter to carry the thermal imaging equipment to assess such a scene costs several thousand dollars per hour. These cannot fly at night or in dangerous weather conditions for safety reasons, and that limits their use.

“But sending out autonomous drones that may be able to fly for $100 an hour/24 hours a day in poor weather conditions – I’m pulling numbers out of the air here – may reduce that cost to the point where we can use them longer and more often, and find more fires more quickly,” he says.

The ANU is testing an Australian manufactured Carbonix Ottanodrone to assess the idea. This 58kg aircraft has a 5.8m fixed wing and can fly up to 100 km/h for a maximum flight time of six hours. It can also take off and land vertically.

But the drone itself is a small part of the equation. Its potential lies in its ability to fly itself while carrying a helpful payload – in this case, a set of thermal and visual cameras with live streaming datalinks.

“We want to demonstrate the ability for the Ottanodrone to take off from a base – in our case, the Spring Valley robotics facility in western Canberra – fly down through the Namadgi National Park to the locations of lightning strikes, get real-time thermal imagery to confirm or hopefully exclude fires – and provide a real-time feed of that data back to emergency services in the ACT,” says Tridgell.

But there are a few technical and regulatory hurdles to jump first.

Drones: out of sight, but in mind

A dry lightning drone must operate safely and effectively over great distances. This is called beyond visual line-of-sight (BVLOS) control.

The ANU is working with Australian drone operations consultancy Hover UAV to produce a comprehensive risk assessment of autonomous aircraft flying over rural areas with a smattering of crewed aircraft nearby.

This will be submitted to the Australian Civil Aviation Safety Authority (CASA) for approval.

Once cleared, Dr Tridgell and his team will put thermal imaging drones through a comprehensive performance analysis.

“We want the fire services to get some hands-on experience using the vehicle so they can make procurement decisions,” he says. “But they can’t decide whether it’s worthwhile unless they’ve got evidence it can do something useful.”

The ANU is testing a smaller drone to monitor hazard reduction burns. It looks much like a normal commercial quadcopter but has a much longer flight time.

“Once we start flying it, we will probably notice problems like perhaps the processing of the thermal imaging isn’t quite right,” says Tridgell. “We want to make sure it only highlights things of real interest – and not just a hot rock under the sun. False positives can be a real problem.”

But so can false negatives.

“It might be that we end up developing some smarter software to give us more useful data under Australian conditions. Or it might be that it works well enough right out of the box. We won’t know until we start trying it,” Tridgell concludes.