I am interested in exploring and developing learning methods that give robots human-like learning abilities. My research focuses on algorithms that allow robots to learn skills from observing humans or other robots and give robots the ability to improve performance while practicing. I am currently investigating the learning of movement skills and the learning of action selection skills while performing dynamic tasks. Examples of tasks that I am exploring include playing games such as tennis and ping pong, performing aerobic movements, and walking. My thesis presents a framework for conducting learning from observation and practice research. The framework is implemented in two tasks; the Labyrinth marble maze game and air hockey. A video of a robot playing the Labyrinth game and a humanoid robot playing air hockey can be seen on my thesis page and more information can be found on my Learning from Observation webpage.
Darrin C. Bentivegna
Displaying 0 Publications