Research
I work on robot learning and contact-rich manipulation, with a focus on how robots can use physical interaction to act reliably under uncertainty.
My research combines probabilistic inference, geometric representations, and sampling-based control. I develop methods that use vision and contact to infer uncertain properties of the environment, then use that information to guide manipulation.
A recurring question in my work is: How can a robot use contact not only to manipulate the world, but also to understand it?
My current research explores this question through contact-aware perception, Bayesian inference, and model-predictive control for fine manipulation tasks such as robotic insertion.