by Sue Palminteri, Mongabay.com
Sharks are in trouble worldwide, with one study estimating that people kill up to 100 million of them each year through illegal fishing, shark finning, and bycatch.
The negative public view of sharks as aggressive killers limits empathy for the ocean’s largest fish, even though only 33 of over 500 species have been known to bite people, and 63 percent of reported attacks have been by three species: tiger, bull, and great white sharks.
Shark-human interactions have increased as people spend more time in the ocean and as more attacks are actually reported. Of the 88 unprovoked shark attacksreported in 2017, five were fatal; to put that into perspective, more people die of lightning strikes than shark attacks.
Limitations of existing shark attack prevention strategies
Previous efforts in Australia to protect surfers and swimmers have relied on nets—which trap not just sharks but also rays, dolphins, and turtles—or drum lines, lengths of baited hooks attached to the seabed that aim to catch and drown passing sharks, as well as direct culling of sharks.
These methods have not proved successful in eliminating shark attacks, and killing the ocean’s top predators, while harming other species in the process, negatively affects marine ecosystems.
More recent non-invasive efforts to detect sharks have included trained shark-spotting teams, electromagnetic devices that attach to surfboards, and expensive seabed-to-surface barriers. Some areas in Australia and Brazil have set out non-lethal “smart” nets and drum lines to capture and tag passing sharks that are then monitored by beacons that receive the “ping” of the animal’s tracking tag.
Each of these methods has had limited success. High-energy surfing zones along Australia’s coast present challenges to maintaining nets and other physical barriers, while tagged sharks may represent a fraction of a total population.
Since 2016, the government of New South Wales, Australia, has been testing a series of new technologies to address the increased number of shark sightings. One of these, a tethered buoy integrating sonar, artificial intelligence, and cellular phone technologies, has shown promise as a non-lethal, non-invasive means of detecting and identifying sharks and alerting officials of their presence near beaches.
Shark-finding technology
The 35-kilogram (77-pound) solar-powered Clever Buoys, developed by Smart Marine Systems (SMS), float on the ocean surface, communicate with devices on shore, and attach to sonar transducers mounted onto a pier or the ocean floor. Each system’s sonar sends out acoustic signals that bounce off objects in the water. It then measures the strength of the returning signal and the time it takes to return to determine the size, distance, and direction of the object.
The sonar reaches objects in the water column up to 120 meters (394 feet) away, so multiple units can be positioned to create a virtual shark net that “sees” underwater, which lifeguards and aerial surveillance cannot. It uses a frequency that is substantially higher than the hearing range of known marine animals, to avoid adverse impacts on wildlife.
The units are set up in an array that is configured according to the form of the beach, SMS founder and executive director Craig Anderson told Mongabay-Wildtech. “Generally it is sitting 400-500 meters [1,310-1,640 feet] from the sand, and then it’s looking out to sea to basically create a virtual barrier or virtual net, and anything that comes within that zone gets identified and analyzed,” he said.
Sharks swim in patterns that differ from those of dolphins, turtles, or rays. The Clever Buoy system runs software called SharkTec that uses AI to identify sharks over 2 meters (6 feet) long based on their movements and distinctive acoustic signature.
“It’s a pattern recognition algorithm of the distinct swimming patterns of various marine life,” Anderson said.
Engineers at SMS and sonar maker Tritech developed SharkTec by adapting software used to detect marine life around oil rigs and tidal turbines with data on shark movements collected over several years.
They “trained” the software by incorporating the shark movement data into the AI algorithms to automate the recognition of sharks through these signature movement patterns; the resulting software distinguishes sharks from other animals by their size, speed, and swim pattern.
When the system determines that a nearby object is likely a shark, it signals lifeguards, who can respond by closing the beach or other action. The system transmits the alert within a few seconds via text message and special mobile phone app with the shark’s location and likelihood that it is a threat. It can also integratethe alert into local emergency services messaging to the public.
SMS is still adapting the system to distinguish among different shark species. Most species mainly avoid people, so the developers and collaborators have also installed the system inside Sydney’s Sea Life Aquarium to test its ability to differentiate among shark species and prevent harmless species from triggering an alarm. A major objective of the NSW Shark Management Strategy is to reduce interactions between beachgoers and sharks while minimizing harm to the area’s marine wildlife.
Testing and retesting
The Clever Buoy developers, together with the NSW government and researchers from the University of Technology Sydney, tested the system, in combination with aerial surveillance, drone surveillance, shark tagging and mobile apps, as part of a larger shark management program to reduce the incidence of shark bites along the NSW coast.
SMS has since tested the buoys at beaches near Sydney and Perth in Australia, as well as at a World Surf League competition in South Africa.
Early tests consisted of placing the Clever Buoy system about 1 kilometer (0.6 miles) offshore, beyond the surf zone, in about 10 meters (33 feet) of water, at well-used beaches. The testing team positioned an underwater video camera nearby to record for up to five hours each day and compared the captured images to the information received from the Clever Buoy to visually verify sharks that the sonar system detected.
The Australian Professional Ocean Lifeguard Association said in a statement that a three-month trial of the system in 2016 at Bondi Beach near Sydney was a success, “with Clever Buoy logging and alerting numerous shark detections, many of which were able to be positively validated through visual identification by the Bondi professional ocean lifeguards.”
The system also successfully identified great white sharks on the NSW central coast, with video footage confirming the detection of the sharks during the trial.
Tests around Perth had mixed results. Local newspaper The West Australian reportedthat two buoy units detected 38 “possible” sharks, which led to 19 beach closures, but that “the system was unable to distinguish between different types of marine animals, let alone species of sharks.”
Anderson said the AI software was a developing technology. “The more time we spend in the water, the more the algorithm develops, the more accurate the system gets,” he said. “It’s a continuum of development, it never stops. Like any pattern recognition development, it’s all about having that library, that catalog of imagery to teach the system and for the system to continually learn.”
The company is now deploying a set of Clever Buoy units at beaches in Southern California. It will lease the system to the city of Newport Beach for between $10,000 and $15,000 per month for six transponders.
The platform has been shown to operate continually in open ocean conditions, autonomously monitoring marine life and alerting authorities when large swimming objects appear to be sharks that may threaten beach users. The Clever Buoy is beginning to carry additional sensors that record environmental indicators, including wave height and swell, water temperature, and levels of certain pollutants and acidity.
This article was originally published on Mongabay.com. It has been republished here with permission under the organization’s Special Reporting Initiatives (SRI) and Global Forest Reporting Network (GFRN) program. To read the original article, click here: https://news.mongabay.com/wildtech/2018/04/ai-buoys-safeguard-swimmers-and-sharks/