Android Academy: Machine learning for robots

To design a robot which can perform tasks with its arms and hands, we can allow robots to learn by themselves through exploration and interaction with the world, in a similar way to how babies learn. "Machine learning" approaches such as this have seen huge success in recent years, particularly with the arrival of "deep learning". In this talk, I will show how we can design robots which can learn hand-eye coordination skills using these ideas from machine learning, with a view towards developing the first commercial domestic robots which can clean and tidy the home.

Robots: the things that nowadays hoover your floor. What does it take to get them to do more useful tasks, like pouring you a pint? Or even better, socialising with you while doing so? In this talk, I take bartender-patron interactions as inspiration for robot-patron interactions. I will focus on a particular human ability - perspective taking - which allows a good bartender to mentally step into your shoes to predict your next order. Finally, we’ll see what robots can teach us about bartenders?