Proposed Sparse Diffusion Policy: a sparse, reusable, and flexible policy for robot learning
Developed Versatile Loco-Manipulation through Flexible Interlimb Coordination
Published work on Singularity-Avoidant Motion Mappings for pose-tracking teleoperation
Invented the Everything-Grasping (EG) Gripper with synergistic suction–grasping capabilities
Created the Prismatic-Bending Transformable (PBT) Joint for modular, foldable manipulators
Research Experience
Engaged in Robotics and Mechatronics research focusing on robotic grasping, dexterous manipulation, human-robot interfaces, and intelligent skill transfer
Developed innovative robotic grippers and manipulators, including Programmable Locking Cells (PLC), Everything-Grasping (EG) Gripper, and Prismatic-Bending Transformable (PBT) Joint
Designed a Dexterous and Compliant (DexCo) Hand based on soft hydraulic actuation and the DIH-Tele teleoperation framework for multi-object manipulation learning
Conducted robot learning research emphasizing multimodal reasoning (vision, force, language), policy representation learning, and continual learning
Worked on Embodied AI and Loco-manipulation, covering teleoperation data curation, generalist pre-training with specialist post-training, and coordinated control across humanoid, quadruped, and arm platforms
Participated in autonomous driving research
Contributed to past projects in intelligent control, mechatronics, and precision motion control, such as exoskeleton design, variable stiffness actuators, HDD control, and MEMS gyroscope calibration