I am a Ph.D. student at The Hong Kong Polytechnic University, advised by Prof. Hongyu Yu. I received my bachelor's degree from the School of Mechatronics Engineering at Harbin Institute of Technology.
My research focuses on robotic manipulation in highly dynamic environments, with policies running directly on onboard edge chips. We are also working on some exciting space robotics projects.
My dream is training large neural nets on massive amounts of robot data.
I was a starting player in the Floorball Asia Cup. Away from robotics, I am also a skilled mahjong and card-game player.
Preprint, 2026 (* equal contribution)
Jointly generates actions and their execution timing in one network evaluation, helping robots manipulate moving objects and complete tasks under time constraints.
Preprint, 2026 (* equal contribution)
Generates actions in just two network evaluations for fast responses to moving targets, including intercepting a moving toy car.
Developed a vision-language-action (VLA) action expert and generative policies for few-step and one-step robotic action generation.
Built a robot simulation and evaluation platform for manipulation and imitation learning, and calibrated and released R1-series URDF models for simulation.