BEFORE NOW, ROBOTS were either programmed to carry out tasks or taught what to do through learning algorithms, but in the future, they may just learn what to do from the internet instead.
Robo Brain is a large-scale computational system that learns from publicly available resources from the internet. The system was developed by researchers at Cornell, Stanford and Brown universities and the University of California, Berkeley.
According to Phys.org, the system is currently downloading and processing about one billion images, 120,000 YouTube videos and 100 million how-to documents and appliance manuals.
This data is being translated and stored in a robot-friendly format that robots will be able to draw upon when they need it. Robo Brain will process images to pick out the objects in them, and by connecting images and videos with text, it will learn to recognise objects and how they’re used, along with human language and behaviour.
This is all done through a process called “structured deep learning” which helps robots classify what objects are and how they’re interacted with. As Phys.org explains:
The system employs what computer scientists call “structured deep learning,” where information is stored in many levels of abstraction. An easy chair is a member of the class of chairs, and going up another level, chairs are furniture. Robo Brain knows that chairs are something you can sit on, but that a human can also sit on a stool, a bench or the lawn.
One of the examples given is that of a coffee mug. If Robo Brain sees this, it can learn that not only is it a mug, but liquids can be poured into it, you can use the handle to hold it and it must be carried upright when full.
Robo Brain will also have teachers, where its website will display things it has learned. Visitors will be able to view this information and make additions and corrections where required, although what they will teach them remains to be seen.
To embed this post, copy the code below on your site
have your say