Chen Yang楊宸

Learning Expressive and Compositional Motion Representation via Spectral Skills

Feiyang Wu, Chenxiao Gao, Chen Yang, Ye Zhao, Bo Dai, Anqi Wu

arXiv preprint2026

A predictive motion representation that enables one humanoid controller to track diverse motions, chain skills, and compose new behaviors by steering in skill space.

EmbodiChain

Core contributor

An end-to-end, GPU-accelerated platform for embodied AI. EmbodiChain brings simulation, automated data generation, and robot learning into a modular workflow for developing and evaluating real-world robotic systems.

Humanoid Locomotion

Reproducing L2T using rsl_rl to train G1 for locomotion across complex terrain. The video demonstrates the learned policy navigating stairs in simulation.

Smart Stop-Snoring Pillow

A soft robotic system integrating pressure sensors, valves, airbags, an air pump, and an embedded controller. It estimates airflow for precise airbag control and uses only pressure signals to detect snoring and estimate heart and respiratory rates.