Every year in Switzerland, roughly 1,000 go missing in the forests and mountains. To combat this problem in the most efficient way possible, a research team at the University of Zurich, Dalle Molle Institute for Artificial Intelligence, and NCCR created rescue drones that can navigate difficult, wild terrain. The goal is that these drones will greatly assist in the search and rescue efforts, dramatically cutting down the search times involved.
The group of small quadcopters can be deployed quickly and follow the man-made trails in forests, which is a lot more difficult than the tree-free environments drones currently navigate.
The software these drones use is a Deep Neural Network, which is a program that mimics the human ability to adapt overall behaviour after learning from a mistake.
In order to prepare the drones for the tough work they’ll carry out, the researchers hiked through the Swiss Alps and took more than 20,000 photos of trails. They used these photos and first hand experiences to create the algorithm. To test the algorithm, they showed the drone’s neural network a trail it had never seen before, and it successfully identified it.
Although a smaller prototype with just two little cameras was used in this test, it was still able to use the algorithm to recognize trails and navigate through them. The most impressive accomplishment from these tests is that the drone prototype was able to recognize a trail 85% of the time, while a human could only do so 82% of the time.
Not only is this research promising for the future of drones, but it could also make a big difference for search and rescue operations all over the world. By cutting down the search time involved, the chances of locating a person safe and alive are dramatically increased. This invention is the latest example of innovation and technological advancement providing a positive addition to the world.
Featured image source: robohub.org