AI learned to control a drone in an unfamiliar space

Researchers at the California Institute of Technology have presented a new way to fly drones in

unfamiliar space. They have developed a machine learning algorithm that allows even multiple devices to autonomously navigate themselves in tight and unfamiliar spaces. The system gives each drone a certain degree of independence, which allows it to adapt to a changing environment.

Instead of relying on cards orroutes of other drones, the new model allows each car to independently navigate in a given space, even if it coordinates its work with others. This decentralized model helps drones to improvise and makes it easier to operate drones, since the computation is distributed among many robots.

An optional tracking controller helps drones compensate for aerodynamic interactions. In preliminary tests, the controller turned out to be more efficient than analogues.

This method can be used when searching andrescue operations where drones can safely sweep areas in flocks, while autonomous devices can minimize traffic jams and collisions. The researchers warn that the method needs to be tested in laboratories, but in a couple of years this model could be commercialized.

See also:

- Look at the huge “wall” of hundreds of thousands of galaxies behind the Milky Way

- Comet NEOWISE is visible in Russia. Where to see her, where to look and how to take a photo

- NASA and ESA publish the most detailed image of the Sun