Po-Chen Ko

I am an undergraduate student at National Taiwan University, advised by Prof. Shao-Hua Sun. I am interested in Computer Vision, Machine Learning, Robotics and Reinforcement Learning.

I was fortunate to visit Josh Tenenbaum’s lab at MIT in 2023 Spring semester, collaborating with Jiayuan Mao and Yilun Du on Learning to Act from Actionless Video through Dense Correspondences.

Email  /  GitHub

profile photo


I am currently exploring video generative model as control policy for robotics.

project image

Learning to Act from Actionless Video through Dense Correspondences

Po-Chen Ko, Jiayuan Mao, Yilun Du, Shao-Hua Sun, Joshua B. Tenenbaum
ICLR 2024 (Spotlight)
arxiv / video / website / code / paper

We developed a robot policy using images, which can be trained without action annotations. It’s trained on RGB videos, effective for table-top tasks and navigation. Our framework also allows rapid modeling with just 4 GPUs in a day.

Design and source code from Jon Barron's website