I am an undergraduate student in Automation at Tongji University, interested in building robots that can perceive, learn, and act in the physical world.
I explore learning-based locomotion and manipulation, connecting algorithms with robotic systems through simulation experiments and hands-on deployment.
Research Interests
I am interested in how robots learn transferable behaviors and connect perception, reasoning, and action across tasks.
Selected Projects
Current Research: nSGS
A state-conditioned neural task sampler for Go2–Piper whole-body control in simulation, with matched-start PPO comparisons and ongoing transfer evaluation.
Fine-tuned π0.5 on 50 teleoperated demonstrations and deployed it on a real RealMan robot for banana pick-and-place, achieving 9/10 successful onsite pilot trials.
