Every year, thousands of people get lost in the forest and mountain areas of Switzerland, with approximately 1,000 calls annually from injured or lost hikers in the country alone. Now, a team of researchers from the University of Zurich may have found the solution to this problem with the development of new software that gives drones the ability to autonomously detect and follow forest paths. The artificial intelligence software has successfully been used to teach a small quadrocopter to search a forest on its own, which could lead to the development of drones that parallel the efforts of rescue teams searching for people in the wild.
"While drones flying at high altitudes are already being used commercially, drones cannot yet fly autonomously in complex environments, such as dense forests. In these environments, any little error may result in a crash, and robots need a powerful brain in order to make sense of the complex world around them," Davide Scaramuzza, who participated in the research, said in a press release.
In order to create a powerful brain, the team used a "deep neural network," which is a computer algorithm that can learn to solve complex tasks and problems by training and being put through example tests, effectively mirroring the process that the human brain goes through when it learns from experience. The researchers used data from hours of hiking along various trails in the Swiss Alps, including more than 20,000 images captured by a helmet camera.
The initial results were promising: when the deep neural network was tested on a novel trail, it was able to find the correct path in 85 percent of the cases, in comparison to humans who could only do so 82 percent of the time. However, more work needs to be done on the software before robots will be prepared to autonomously scour the forests.
"Many technological issues must be overcome before the most ambitious applications can become a reality. But small flying robots are incredibly versatile, and the field is advancing at an unseen pace. One day robots will work side by side with human rescuers to make our lives safer," said Luca Gambardella, a member of the research team.
"Now that our drones have learned to recognize and follow forest trails, we must teach them to recognize humans," added Scaramuzza.
The findings were published in the Dec. 17 issue of IEEE Transactions on Control Systems Technology.
