Bleeding Edge: Researchers at the University of Zurich (UZH) have developed a machine-learning algorithm to control a quadcopter drone that can outperform professional racing drone pilots. This algorithm calculates the "optimal time paths" while calculating the limits of the drone.
This masterpiece seems obvious at first glance - the human machine learning system is on the buzz again, so what? However, drone enthusiasts are very prominent at what they do, and this is the first time an autonomous system has outpaced world-class human pilots.
To test the system, UZH researchers conducted an unmanned flight course (below). Both drones and human pilots were allowed to train on the road. The AI was not only able to achieve the fastest time in the distance, but also due to the large margins it was able to pass professional pilots from anywhere.
AI uses external cameras to track the drone's path and make appropriate calculations. . The team hopes to upgrade the system with quad cameras. Using CPU camera systems to perform other tasks related to drones is critical. The researchers expect their work to be useful for applications such as search and rescue, building inspection, package delivery, and more.
Algorithm "An algorithm is also a creditor". The computer now takes up to an hour to accurately calculate the optimal path. Because of this shortcoming, at least for the time being, human pilots have no alternative out of fear. Obviously in situations like search and rescue, when time is of the essence, they want a program that can more quickly calculate its path through waypoints.
All technical details are described in the team's recently published article in Robotics Science.
Image Credit: University of Zurich p> p
This fully autonomous drone outperforms the two global competitors of human drones