Welcome to my home page!
NEWS: The XML files used for the ICRA submission are available here.
I am a doctoral student at the Media and Machines lab, where I am working with Bill Smart.
I study motor control; in particular, how optimization can be used to generate novel motion from first principles. I seek ways to train my simulated agents to perform complex, articulated motor tasks, often resorting to multi-stage training regimes as a way to shape a certain behavior.
The tools I use come under a variety of titles: optimal control, reinforcement learning, and adaptive dynamic programming. I focus on continuous, high-dimensional, non-linear domains, and so my research tends towards local methods. Most of my work is done in collaboration with Yuval Tassa.
Here are some of our recent demos:
Footed Walkers
Swimmers
This is a 34-dimensional system, and none of the exhibited behaviors (swimming, turning, coasting, and braking) were pre-programmed.
7DOF arm
This is an OpenSim model of a 7DOF arm balancing a pole. A feedback policy was learned first, and then tested by the perturbations (illustrated with a green arrow).
Walkers
A comparison between the initial walker and a walker on stilts:
The composite controller traversing rough terrain: