Some of us happen to live in an environment where mistakes are not tolerated at all, and this is where most folks fail to grow or learn from the process. Mistakes are there for a reason – and a very good one, to boot. After all, it helps us figure out what not to do in the future, and hence, to err is human. Well, how about building a robot that also walks in a similar manner, that is, to be able to learn new skills simply by being a human being?
After all, one can say that a robot is as smart as its programming, and learning has always been a process of the living. A bunch of scientists at UC Berkeley has successfully programmed a robot to have it learn basic tasks through trial and error, in the same way as that of a human. The robot itself is nothing new, as we have seen how a Willow Garage PR-2 has worked in the past in different situations, but being equipped with the new Deep Learning artificial intelligence, you can say that the PR-2 has “gained” a select brand of primitive learning ability.
This robot has been given the nickname of BRETT, (Berkeley Robot for the Elimination of Tedious Tasks), where it will make use of both visual and sensory information concerning itself, its environment and the objects located right before it, building them up like LEGO and figuring out how things work. Pretty neat, don’t you think so?
Filed in . Read more about AI (Artificial Intelligence) and Science.