Scientists at Cornell University are working on a repository of robot knowledge which can process data from YouTube videos, images and manuals about human behaviour via the internet, intended for sharing among similar machines globally.
According to scientists, the new computational system dubbed ‘Robo Brain’ is currently downloading and processing about a billion images; 120,000 videos from YouTube; 100 million how-to documents and appliance manuals; in addition to earlier training offered to other robots.
Upon being trained at these materials, the Robo Brain will be capable of making out objects and their usage, in addition to human language and behaviour and share this knoledge among other robots.
Cornell University computer science assistant professor and Robo Brain lead researcher Ashutosh Saxena said: "Our laptops and cell phones have access to all the information we want.
"If a robot encounters a situation it hasn’t seen before it can query Robo Brain in the cloud.
"The information is being translated and stored in a robot-friendly format that robots will be able to draw on when they need it."
The downloaded data will be stored in a Markov model, involving a series of points (nodes) linked by lines (edges), such as a massive branching graph, that facilitate robots to draw on when required.
Backed by Google, Qualcomm, and Microsoft among several others, the Robo Brain is mainly aimed at enabling more free interaction among robots and the world, with applications ranging from home appliances to cars.
