In this episode of Judah vs. the Machines, Judah visits UC Berkeley’s Robot Learning Lab to meet Brett, a robot still in testing that was designed to learn like a human. Brett’s creators hope that he could some day take on menial human tasks with the ability to adapt and learn new ways of accomplishing a goal, something that most other robots are incapable of doing.
In the field of machine learning, this type of learning is known as reinforcement learning. It’s pretty popular these days because it holds the potential to make machine learning less rigid and breakable.
Like a human, Brett takes actions in order to maximize his reward. The catch is that he doesn’t know which actions will lead to the best outcome. Adopting a trial-and-error approach, Brett is anything but fast, but he is able to solve tasks with less background knowledge than his peers.
In order for Judah to compete against Brett, the research team organizing the skirmish had to handicap him — giving Judah thick mittens and a blindfold to take his senses out of the equation. With the match set up, Judah pours his heart into his toughest challenge yet.
See all eight episodes of Judah vs. the Machines here.