During the past few months, news about drones and their abilities have been countless. Commercially used drones fly at a relatively high altitude where they hardly encounter any obstacles. However, a team of Swiss researchers from the Dalle Molle Institute for Artificial Intelligence, the University of Zurich and NCCR Robotics has now developed an Artificial Intelligence software which allows quadrocopters to navigate autonomously at lower altitudes, mainly in forests. Using a Deep Neural Network (DNN), the researchers taught the drone to recognize forest trails and follow them. A DNN works similarly to a human brain as that it is learning from experience: A computer algorithm learns from examples to solve complex tasks. In order to gather these examples, the researchers took more than twenty thousand pictures of hiking trails using helmet cameras. After ‘teaching’ the computer on the drone, the quad-copter was able to follow forest trails autonomously even if they had not been part of the picture series.
The research has been published in IEEE Robotics and Automation Letters and will be presented in May at the IEEE International Conference on Robotics and Automation in Stockholm. Finally, the project is aimed at producing drones which can be used to rescue people who went missing in the wild.