Strategy Big Data Battling extinction with drones and AI-powered analytics

Battling extinction with drones and AI-powered analytics

Beverley Head — September 2018

How Fujitsu’s Digital Owl project has used cutting-edge technology to help protect threatened species in the Australian bush.

Extinction is forever. So with around 1,000 plants and animals at risk of extinction in the Australian state of New South Wales (NSW), there’s a concerted effort underway to tackle the problem.

The NSW Government has already committed A$100 million ($72m) over five years to address the challenge. While that is an impressive commitment, there’s a vast amount of ground to cover — quite literally — and limited resources in terms of on-the-ground rangers.

To stand a chance of success the government needs to get maximum bang for its buck. In a recent trial, technology first developed for body-mounted cameras worn by the State’s police officers has been adapted to monitor at-risk plants and invasive weeds potentially threatening their habitat.

In a trial that combines drones, cameras and artificial intelligence, Fujitsu and the NSW Office of Environment and Heritage have demonstrated how innovative solutions can amplify the fight against extinction.
Eye in the sky
Lee Stewart, head of sustainability for Fujitsu in Australia and New Zealand realized five years ago that advanced technology could be a conservation ally. He’d asked park rangers about the major source of the CO₂ emissions that they were responsible for generating and was surprised to learn it was helicopter fuel from their forays to identify and protect at-risk plants.

Lee Stewart, head of sustainability, Fujitsu Australia and New Zealand

Helicopters would be hired to fly over bushland and tag the presence of rare plants or invasive species, using GPS. In some cases they would winch down a ranger with a chainsaw to take out a feral tree; in others, teams would be sent in later to either protect or eradicate plants. Given that NSW has a land area of more than 800,000km², the cost of using helicopters in such tasks was considerable, let alone the contribution to emissions.

Stewart pitched the idea to Fujitsu Oceania’s Incubator program which funds innovative co-creation projects, and secured finance for a trial. The Digital Owl project was about to take flight.
Saving species
To get to Mount Dangar you drive north from Sydney to Sandy Hollow, and turn off before Giants Creek. Then you walk for a day.

It’s spectacular but remote country – and the only place in the world where Acacia dangarensis grows. Mount Dangar is also home to a rare daisy – and, like many parts of NSW, it faces the scourge of the non-indigenous prickly pear, a noxious weed that threatens bio-diversity in many areas of Australia.

Working with a team from the NSW Office of Environment and Heritage’s Save our Species program, Stewart led a team that included specialists from drone-maker Carbonix and Fujitsu developers to create the Digital Owl trial that showed how technology could be harnessed to help protect threatened species and support rangers planning evidence-based strategic action.

The system has enormous potential, according to Lucas Grenadier, senior team leader for ecosystems and threatened species, for the Hunter division of the Office of Environment and Heritage. And it’s not just Mount Dangar facing the extinction challenge.

As NSW Environment Minister, Gabrielle Upton, notes: “In NSW alone there are approximately 1,000 plant and animal species under threat of extinction. Saving these species is crucial to the ongoing health of the various ecosystems in NSW. However, monitoring such a broad area can be prohibitively expensive, especially when considering the cost of chartering and fueling helicopters. It’s exciting to be using new drone technology with detailed layers of analytics behind them to get more accurate information including maps of otherwise inaccessible areas.”
Learning machine
Working with Fujitsu developers in the UK and Japan, Stewart’s team leveraged Fujitsu’s Advanced Image Recognition, which is powered by the company’s Zinrai artificial intelligence platform, and can be deployed in the cloud for processing-intensive analysis and interpretation of images. During the trial, around 5,000 images captured by the drone were analyzed.

Scanning for endangered species over Mount Dangar, New South Wales

Stewart says that the recognition engine is already working at accuracy levels of 70%, and further refinement of algorithms should push that above 80% in coming months. By optimizing the image recognition system, “we can fly higher and faster, and cover more ground,” he says, with no loss of accuracy in terms of the ability to identify species.

Digital Owl’s first trial exceeded all expectations. The algorithms were able to identify and map the presence of acacias and prickly pears – as well as a rare type of daisy that hadn’t been observed in the wild for more than a year. To make doubly sure of the discovery, and to ensure it wasn’t a false positive, a ranger was dispatched to the exact location on foot and confirmed the reading by Digital Owl.

Capturing data for analysis by Fujitsu Advanced Image Recognition: Mount Dangar, New South Wales

Stewart believes there’s an opportunity to use Digital Owl to provide important information to conservationists about the impact of drought on endangered plant species and threatened animal populations. Periodic surveys using the system would provide important longitudinal data that can help rangers plan more strategic protection or eradication efforts, ensuring maximum impact.
Soaring possibilities
Stewart says the applications for a Digital Owl-like solution are extensive. The next step is to integrate it with Fujitsu’s SPATIOWL big data and cloud platform, to extend its capabilities and open up the option of real-time data transmissions from the drone, uploaded to the internet via satellite. That would ensure anytime, anywhere data analytics.

But the scope is not restricted to species on the verge of extinction. “We’re also looking to expand Digital Owl into finding koalas, monitor the wild buffalo causing problems in the Northern Territory and validating carbon-farming programs.” He also foresees a range of applications that might include search and rescue, surf life-saving and other emergency services, where data can be collected remotely from the air and instantly analyzed to inform decision-making. “There are a lot of opportunities and I’m confident that by the end of the year we will have a commercial project,” he says.

Meanwhile, there are acacias and daisies growing on Mount Dangar that now have a more secure future — and some rightly worried prickly pears.

• Watch a video on the DigitalOwl project

First published September 2018