Saving Africa’s Last Rhinos

You can tell a lot about a poacher by the way they dehorn a rhinoceros.

Was the horn hacked off crudely — snout and all — with a machete? Or was it removed skillfully with a sharp, scalpel-like instrument, right along the thin partition of cartilage that separates horn from bone?

As Saif Bhatti bumped along the dirt back roads of Thornybush Game Reserve in South Africa, he was unsure which one they might find. He sat shotgun in an old pickup truck next to the reserve’s security manager, who held his phone in his right hand, gripped his radio in his left and steered through the savanna with his elbows. Someone in the area had reported two gunshots fired around 6 a.m. But in the sprawling 35,000-acre park, it was hard to know where to search.

With more than 8,000 poached in South Africa from 2010 through 2019, rhinos are likely to become extinct within our lifetime. Habitat loss is partially responsible for the rhinos’ dwindling numbers. But, overwhelmingly, poaching remains the largest threat. On the black market rhino horn is worth more than gold, which fuels illegal hunting.

The search for the poachers in Thornybush took place throughout the day and night and included helicopter, patrol and canine units at different times.

It wasn’t until the next morning that the security manager’s radio crackled to life: His team had discovered the carcass of the reserve’s last female black rhino.

The first bullet had ripped through the right side of the rhino’s abdomen. As she turned to run, a second shot blew through her left shoulder. She finally stumbled to the ground and — likely painfully and slowly — bled to death. The rhino was pregnant; the fetus did not survive.

Where the rhino’s horns once perched upright, now there were exposed sinuses, glistening and yawning toward the sun.

“The horn had been expertly removed,” Bhatti says. “It was not this group’s first poach.”

But if Bhatti’s new technology works as well as he imagines, this poach could be one of their last.

Purpose-Built Technology

Bhatti’s device uses signal processing to distinguish gunshots from other loud noises, such as thunder, branches breaking or elephants stampeding. To refine gunshot detection, Bhatti has worked closely with Stephen Tarzia ’09 MS, ’11 PhD, an assistant professor of computer science in the McCormick School of Engineering. An expert in acoustic sensing, Tarzia and his graduate students, Jiayue Sheng and Yuxing Guo, have helped Bhatti implement signal processing algorithms into his devices.

The basic challenge is to make a computer capable of listening for gunshots, Tarzia says. “The software must be simple enough to run on a cheap, solar-powered microprocessor but also sophisticated enough to distinguish between a gunshot and similar sounds.”

To train the device, Bhatti and his collaborators went to an outdoor shooting range. They recorded sounds from guns commonly used in poaching — including those with silencers.

The shooting range trials paid off. “In tests at Thornybush during the African winter, our device was able to detect a nonsuppressed weapon from 800 meters away and a suppressed weapon from 300 meters away,” Bhatti says. “Based on those estimates, we calculated that just 20 units could cover an area of 10,000 acres or so.”

Once a Renoster unit registers a gunshot, it sends relevant data to the cloud for trilateration, the process of determining locations by measurement of distances. Because cell service can be unreliable in the bush, Bhatti is focused on building a mesh network based on radio communication. For power, the device contains a battery that recharges with solar energy.

Africa has implemented other technological strategies to combat poaching, but none were specifically designed and built with poaching in mind, according to Bhatti. 

“There are so many small intricacies and engineering challenges in the savanna,” Bhatti says. “It’s extremely hot with very little power supply and lots of dust. We wanted to make a purpose-built technology that takes those things into account.”

Tarzia credits Renoster’s innovative technology to Bhatti’s visionary thinking and collaborative approach.

“Saif has excelled at creating a vision for his project, forging partnerships and building prototypes,” Tarzia says. “My main role has been to slow him down when necessary to show him how to navigate the engineering product development steps that lead from that initial idea to successful deployment.”

Animated Infographic: How Renoster Works; Art: Matt Twombly

Try, Try Again 

Of course, Bhatti’s system also has faced its own challenges.

Last summer he mounted 15 Renoster units in trees across Thornybush. From there, he encountered multiple unanticipated engineering challenges. Everything from the types of trees to the heat to the rain disrupted his plans.

“I wanted to put the devices up in the trees to avoid interference from the deviations of sound on the ground,” he says. “But then I learned that elephants just like smashing trees. It’s a favorite hobby of theirs. I had to make sure to pick sturdy trees, like the marula.”

And then there was the weather.

During the summer rainy season, parts of South Africa receive heavy amounts of rain. A big challenge is to make the units porous enough to allow sound to enter them while ensuring that they remain water- and dust-proof. Bhatti first put boxes around the devices to waterproof them. The boxes, however, made the devices more susceptible to heat. “It turns out that when you put a device in a black box out in the summer sun, it’ll bake,” Bhatti laughs. For his next iteration, he used a special UV coating.

To help solve the waterproofing problem, Bhatti used Gore-Tex material to make a semipermeable acoustic vent that would prevent dust and water droplets from getting inside the device but still allow the microphones to pick up sound.

In late December, Bhatti returned to Thornybush Game Reserve to add the acoustic vents to his devices. He also collected data from the devices to see how well they processed and correctly identified sounds.

“When I first installed the devices, it was winter, so the bush was dead,” Bhatti says. “When I returned in the summer, the landscape was lush with greenery. The additional foliage really influences how sound dissipates away from a gunshot, so capturing that difference was important.”

Those working with Bhatti expect that he will press on until he perfects Renoster.

“Saif’s primary personality trait is persistence,” Niewoudt says. “He simply does not give up.”