Drones and machine learning form shark safety system
Drones and machine learning form shark safety system

Drones and machine learning form shark safety system

The University of Technology Sydney and drone specialist The Little Ripper Group have developed an aerial shark safety system to protect beach-goers from deadly attacks. 

Australia has had its fair share of deadly shark attacks in recent years. The safety of swimmers and surfers is of such concern that the New South Wales government launched a $16 million Shark Management Strategy back in 2015. Part of that money has gone towards innovative trials of drones as a means to detect sharks from above and provide an early warning to swimmers.

After a successful pilot program last year that saw UAV company Little Ripper deploy drones and beam a live feed back to an onshore team, the shark safety system is going to be put into practice on some of Queensland’s most popular beaches from September.

What lurks beneath

Spotting sharks under the surface is a near-impossible task when you’re swimming or surfing. In a bid to keep people safe on the coast of Australia, researchers have developed an aerial system capable of flying autonomously over dangerous coastlines to identify sharks from above.

As well as helping to detect sharks, the Little Ripper Life Saver drones will also be able to deploy a small inflatable and other potentially lifesaving gear to swimmers in need of assistance.

Read moreOcean Alliance deploys drones to gather whale data


Spotting sharks from above

Although a bird’s-eye view offers a safer, more useful position from which to spot sharks, people still make mistakes when looking through footage in search of tell-tale fins. At best, they’ll accurately pinpoint sharks around 30 percent of the time, according to Dr Nabin Sharma, a research associate at the University of Technology Sydney’s School of Software.

The new detection software can boost the success rate to closer to 90 percent. Machine learning algorithms have been trained to spot sharks from both moving and still images. The system is also capable of distinguishing sharks from other animals in the water.

Little Ripper’s shark safety system still requires human supervision to verify the results. “It’s not about replacing human beings altogether,” said Sharma. “It’s about assisting human beings to get the work done in a better way with more accuracy. That’s what the application is meant for.”

Read moreAustralia-based WaterGroup partners with Thinxtra on smart water monitoring

Adaptable shark safety

In response to growing numbers of shark attacks off the north-eastern coast of Australia, some controversial steps have been taken by local authorities in recent years. These have included the use of shark nets, a brute-force solution unable to distinguish dangerous predators from other, potentially rare marine life.

Professor Michael Blumenstein of University Technology Sydney (UTS) has been leading the software development project. He said that by using “cutting edge deep neural networks and images processing techniques…the system efficiently distinguishes and identifies sharks from other targets by processing video feeds that are dynamic as well as images, where objects are static.”

“The system will be able to warn swimmers/surfers from overhead with an onboard megaphone attached to Westpac Little Ripper drones when a shark or a potential risk is detected. This cutting-edge AI system developed by UTS will create a positive impact for the public while making beach recreation much safer and more secure.”

Read more: Underwater NB-IoT: SMRU tracks seals for environmental data