|
Abstract:
The robot learning research has garnered significant interest over the past few years. With the great success of Large Language Models, the robot learning community has been wondering when we will make the scientific breakthrough, just like those made in the language and vision community. However, multiple challenges still need to be addressed.
In this talk, I will first present my recent work on building a state-of-the-art robot manipulation policy using diffusion modeling and 3D scene representations. I will then discuss the challenges I have confronted, including distributional shift, physical understanding and progress monitoring. The goal of this talk is to help the listeners better understand what are the limitations of robot learning and encourage participation in robotic research.
Biography:
Prof. Tsung-Wei Ke received Ph.D. (2022) from University of California, Berkeley. He has spent two years as a postdoctoral researcher at the Machine Learning Department of Carnegie Mellon University. Currently, he is an assistant professor at the Computer Science and Information Engineering Department of National Taiwan University, leading the Embodied Artificial Intelligence Lab.