Artificial intelligence algorithms continue to perform better than humans in certain tasks. Now they can beat a person not only in the game of Go, but also in the drone race.

Robotics specialists from the University of Zurich have developed a neural network capable of controlling a quadcopter to defeat a human operator. Earlier, these same researchers showed algorithms for participation in the race, but the first time they lost to people. Their problem was that they built the trajectory of the drone from a set of smooth curved lines. The man, on the other hand, acted more sharply.

The new program was able to operate the device almost like a human. The nuance is that while the calculations for controlling the drones have to be done in advance. Moreover, these are resource-intensive calculations. Several minutes of flight have to be calculated for more than an hour. However, the researchers hope to improve the program and make it “think” in real time.

Read also:

And here is the newest smartphone with record battery life. It is unlikely to be cooler in 2021

Doctors told how to determine dangerous blood sugar levels in the palms

The doctor told what to eat at night to lose weight in a dream

The flagship iPhone 12 Pro Max fell record price

If you were looking for an alternative to “impudent Chinese with more expensive smartphones”, you just found it