Self-driving drone with a maximum speed of 40 km / h Recognize obstacles even in unknown environments and fly: Innovative Tech
Self-driving drone with a maximum speed of 40 km / h Recognize obstacles even in unknown environments and fly: Innovative Tech
By uavtechnology
08 May 22
This section is written by Hiroki Yamashita, who presides over the web media "Seamless" that introduces the latest research in technology. Mr. Yamashita picks up and explains highly novel scientific papers.
Developed by the University of Zurich in Switzerland and a research team at Intel, "Learning high-speed flight in the wild" is a drone automatic that can break through the environment at high speed without hitting obstacles in a completely unknown environment that you have never seen. It's a driving system.
Drone flying in an unknown environment Trajectory of a drone flying in an unknown environment
When exploring complex and unknown environments such as forests, buildings and caves, fast, agile and small drones are efficient, but autonomous drones cannot navigate the unknown without a map.
The research team trains autonomous drones to fly in previously unseen environments and proposes a method to fly to a destination without colliding with miscellaneous obstacles while maintaining a speed of up to 40 km / h. .. All this is achieved using only the camera and calculations on board the drone.
Predict the optimum trajectory based only on the data from the sensor using CNN (Convolutional Neural Network). This one-step method can be processed faster than the conventional two-step system, which creates a map of the environment from sensor data and then plans the trajectory in the map. As a result, since the judgment is quick as appropriate, it is possible to avoid obstacles at high speed and pass through in an efficient orbit.
System overview
The approach could help improve the performance of self-driving cars as well as drones, he said. The research team also explains that it may even open the door to new ways to train machine learning systems for operations in environments where data collection is difficult or impossible, such as on other planets.
Source and Image Credits: Antonio Loquercio, Elia Kaufmann, René Ranftl, Matthias Müller, Vladlen Koltun, and Davide Scaramuzza, “Learning high-speed flight in the wild” SCIENCE ROBOTICS, 6 Oct 2021, Vol 6, Issue 59, DOI: 10.1126 /scirobotics.abg5810
