Every year hundreds of thousands of people worldwide get lost in the wild. In Switzerland, around 1,000 emergency calls per year come from hikers, most of whom are injured or have lost their way. Drones are an efficient complement to human rescuers and can be deployed in large numbers, are inexpensive and prompt, and thus minimize the response time and the risk of injury for those who are lost and those who work in rescue teams.
A group of Swiss researchers from the Dalle Molle Institute for Artificial Intelligence, the University of Zurich and NCCR Robotics has developed artificial intelligence (AI) to teach a small quadrocopter to recognize and follow forest trails by itself. This research is a premiere in the fields of AI and robotics and could soon be used in parallel with rescue teams to search for people lost in the wild faster than would be achievable by human rescuers.
“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,” says Professor Davide Scaramuzza from the University of Zurich.
The drone used by the Swiss researchers observes the environment through a pair of small cameras, similar to those in your smartphone. Instead of relying on sophisticated sensors, their drone uses very powerful AI algorithms to interpret the images to recognize man-made trails. If a trail is visible, the software steers the drone in the corresponding direction.
Credit: NCCR Robotics
“Interpreting an image taken in a complex environment such as a forest is incredibly difficult for a computer; sometimes even humans struggle to find out where the trail is,” says Dr. Alessandro Giusti from the Dalle Molle Institute for Artificial Intelligence.
The Swiss team solved the problem using a so-called Deep Neural Network, a computer algorithm that learns to solve complex tasks from a set of “training examples,” much like a brain learns from experience. To gather enough data to “train” their algorithms, the team hiked several hours along different trails in the Swiss Alps and took more than 20,000 images of trails using cameras attached to a helmet. The effort paid off: when tested on a new, previously-unseen trail, the deep neural network was able to find the correct direction in 85 percent of cases; in comparison, humans faced with the same task guessed correctly 82 percent of the time.
“In the last eight years we have developed huge Deep Neural Networks (DNN) to solve difficult problems from the fields of biology, automation and document processing,” says Dr. Dan Ciresan or the Dalle Molle Institute for Artificial Intelligence. “This is our first attempt at creating a small but performant DNN capable of running on a computer on our drone. I am happy to see that the same networks we have used to analyze biological brains, detect cancerous cells and diagnose retinal disorders can also be used to drive autonomous quadcopters.”
The research team warns that much work is still needed before a fully autonomous fleet will be able to swarm forests in search of missing people.
“Many technological issues must be overcome before the most ambitious applications can become a reality,” says Prof. Luca Maria Gambardella, director of the Dalle Molle Institute for Artificial Intelligence. “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: this is a small but important step in that direction.”
“Now that our drones have learned to recognize and follow forest trails, our next step will be teach them to recognize humans,” says Scaramuzza.