Saving Rudolph

Boulder-based company uses artificial intelligence to protect Norwegian reindeer

Like Facebook recognizing the faces of your friends when you tag a picture, ReindeerCam is capable of detecting reindeer, ultimately saving them from collisions with trains.
Courtesy of ENGINE27

Not too far from the North Pole, Norwegian trains have killed more than 100 reindeer in a recent three-day period. Train companies claim the deaths occurred because train engineers didn’t receive warning messages about nearby herds. Often, reindeer struck by a train don’t die instantly and the concentrated rate of collisions has sparked outcry from environmentalist groups and reindeer herders seeking more protection for reindeer.

Boulder AI, a local engineering company, believes the best way to address this problem is through artificial intelligence (AI), machines capable of performing tasks without human intervention. After partnering with ENGINE27, a Belgian tech company, Boulder AI developed ReindeerCam, a smart camera capable of detecting reindeer.

After hearing about Norway’s situation with the reindeer, Darren Odom, CEO of Boulder AI, contacted the railroad companies.

“Our goal was to show that the problem is totally solvable,” he says.

Odom began his work with AI as a contractor with the U.S. Geological Survey and U.S. Bureau of Reclamation to develop technology that could recognize various fish species. He realized artificial networks could improve traditional technology and from there, a simple mathematical model turned into something of futuristic potential.

For ReindeerCam, he realized he needed a camera that could operate in extreme weather and function off-grid.

“I noticed that there was a need to have a camera that can do processing at the ‘edge,’ away from high-speed internet connections,” Odom says.

ReindeerCam operates as a solar-powered system in which all cameras in the network communicate over a radio network. An algorithm developed by ENGINE27 gives the cameras the ability to recognize a reindeer and send an alert to the train engineer to slow down.

Geert Van Herbruggen, the CEO at ENGINE27, says developing the algorithm was simply a matter of collecting photos of reindeer from the internet and programming them into the system.

Think of it like Facebook recognizing the faces of your friends when you tag a picture. ReindeerCam incorporates the same technology to give the camera the information it needs to determine whether a reindeer is in a train’s path.

According to the Harvard Business Review, image recognition has become one of the fastest growing innovations in AI software. The margin of error for recognizing images from a large database went from more than 30 percent in 2010 to 4 percent in 2016.

While typical security cameras require someone to manually look through footage, image recognition eliminates the middle man, making for a faster and more efficient system.

On top of saving reindeer, AI software can advance small business models. This is where Boulder AI has made its stake in this technology.    

After seven months, Odom and his colleagues have developed AI technology for a bank, a pizza restaurant and nature conservation groups with each of its cameras tailored for individual clients.

“The core business that we are focusing on is vision as a service,” Odom says. “If we find a company to partner with that has a specific need, we are able to go into the particular market and apply AI to challenging problems.”

Denver-based coffee business Bext360 uses Boulder AI’s algorithms to mechanize payment through the supply chain and better collect and analyze its product. Daniel Jones, CEO of Bext360, says the ability of AI software to operate in challenging environments has helped his company the most.

Despite competing against big tech companies like Google and IBM, the modest size of Boulder AI gives Odom and his team the chance to ensure their work is competitive and that they stay on top of the latest technology.

While detecting reindeer may seem low in the grand scheme of technological innovations, it demonstrates the power of AI technology. With companies like Boulder AI and ENGINE27, integrating this technology into everyday life is much more attainable.

In hopes of saving one reindeer at a time, Boulder AI, with ENGINE27’s algorithm, will work to establish ReindeerCam’s pilot study in Norway. From there, the possibilities are endless as these companies look to grow with the rapid developments in AI technology.