FlowDPG — AirPods assembly.
Kexin Shi 石可心 — to one's heart's content
I build online learning systems where robots explore on their own to acquire capable, reliable, and generalizable behavior in physical world.
I am a second-year Robotics Ph.D. student at the Carnegie Mellon University Robotics Institute, advised by Prof. Deepak Pathak. Meanwhile, I am also a student researcher at Skild AI. Previously, I was a graduate researcher at the Robotics and Perception Group, advised by Prof. Davide Scaramuzza.
NEWS
FlowDPG was invited to a spotlight talk on RSS 2026 Diff4RL workshop .
Extreme Parkour was accepted to ICRA 2024. See you in Yokohama!
Invited talks on Extreme Parkour, at Stanford REAL, MIT Biomimetic Robotics Lab, and the Oxford Robotics Institute.
Agile Flight was accepted to ICRA 2023. See you in London!
PUBLICATIONS
* denotes equal contribution † denotes equal advising
FlowDPG: Deterministic Policy Gradient on Flow Matching Policies for Real-World Manipulation
Extreme Parkour with Legged Robots
Life is an adventure.
Let's keep exploring! ♡