Each year thousands of hikers end up lost in the woods. The search and rescue missions prompted by such occurrences require manpower and resources. Researchers at the Univ. of Zurich’s Dalle Molle Institute for Artificial Intelligence are developing a drone capable of autonomously searching forest and mountain trails for the injured and lost.

“While drones flying at high altitudes are already being used commercially, drones cannot yet fly in complex environments,” said Prof. Davide Scaramuzza. “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.”

In order to impart this navigational knowledge to drones, the research team spent hours hiking along different trails in the Swiss Alps, photographing over 20,000 images from cameras attached to helmets. These images were then used to design a computer algorithm, which the researchers call the “Deep Neural Network.”

The researchers’ laboratory has been researching deep learning and artificial intelligence since the 1990s. And their efforts have paid off. In tests, the drone, which glimpsed its surroundings using cameras similar to the ones found in modern smartphones, successfully identified directions on a path it hadn’t previously seen 85% of the time. Conversely, humans performed the same task with 82% accuracy.

“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 the human rescuers to make our lives safer,” said Scaramuzza. “Now that our drones have learned to recognize and follow forest trails, we must teach them to recognize humans.”

Robots working in the search and rescue field isn’t a new idea. Already, researchers are testing the feasibility of using snake-arm robots to assist with search and rescue missions.