Drones and artificial intelligence are a powerful combination. They can already be used to kill, but in the near future they might also be used to save you when you take a relaxing hike and end up getting completely lost.
Researchers from several universities in Switzerland published a whitepaper last week that explains how "deep learning" technology can empower drones to find and follow hidden trails with more accuracy than human rescue teams.
The team hiked several hours along different trails in the Swiss Alps and took more than 20,000 images using cameras attached to a helmet, according to a news release. Those images were then used to develop algorithms to train drones how to follow any similar trail, even ones they hadn't seen before.
When tested on a new trail, the researchers' algorithm was able to find the correct direction in 85 percent of cases. Humans were only able to find the trail 82 percent of the time.
The program works without any specialized sensors. Because the drone flies just a few feet above the trail, it needs only two simple, smartphone-type cameras to scan the trail, rather than more expensive and less-reliable heat-imaging sensors that can penetrate a forest canopy from above.
The software is the key to the drones' success. "Interpreting an image taken in a complex environment such as a forest is incredibly difficult for a computer," said Dr. Alessandro Giusti from the Dalle Molle Institute for Artificial Intelligence. The deep-learning technique his team developed is based on Deep Neural Networks, a technique that analyzes every single pixel of an image to find a pattern.
But software alone isn't the only thing that needs to be perfected if the drones are ever to be used in wilderness rescues. Apparently, drones are targets for birds of prey, so in order to save lost hikers, the drones themselves have to be protected from predators who may think they look tasty.
This article originally appeared on PCMag.com.