[an error occurred while processing the directive]
RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

Daisy (robot of Facebook)

Product
Developers: Facebook
Date of the premiere of the system: May, 2019
Technology: Robotics

Content

2019: Announcement

On May 20, 2019 Facebook showed the first hardware robot —  Daisy. It is similar to a big spider, and the company teaches it to learn the world by means of artificial intelligence, using a trial and error method.

By the time of demonstration of the six-legged robot researchers taught him to interact with controllers that allowed it to move. After that it was succeeded to develop a certain similarity of tactile sense at the device that helped the robot to roll a ball independently.   

The Daisy robot developed by Facebook
File:Aquote1.png
The real world is difficult  — Roberto Calandra, the research associate of department of artificial intelligence  of Facebook says.  — It  is not ideal,  is not transparent. Therefore the fact that we try to develop the algorithms working on  real robots will help to create more reliable AI systems capable to study quicker.
File:Aquote2.png

Facebook is engaged in robotics within laboratory of artificial intelligence Facebook AI Research (FAIR). Her employees aim at that AI systems studied independently at a basis of small data sets.

To give to robots the chance to samoobuchatsya, scientists applied predictive model of deep learning which was initially developed for the analysis of video. This technology, making a start from the current frame, tries to predict future succession of events. In a case with robots this model helps them to understand how it is better to work in this or that situation.

In development of robotics Facebook pays attention to three main directions: to cognition of the world using contacts, using curiosity and to training in the natural way on the basis of the analysis of experience.

In the first case the company developed a method which allows robots to study by means of touches and on the basis of it already to carry out tasks, without having at the same time any ready information. 

In the second case an AI system encouraged every time when it tried something new. Researchers consider that the curiosity is a good catalyst for training of robots.[1]

Robotics



Notes