Papers
11
Total Citations
321
H-Index
7
About
Junhang Wei is a leading researcher in robotic dexterous manipulation, with a focus on visuotactile sensing, multimodal perception, and skill transfer. His work addresses fundamental challenges in enabling robots to perceive and interact with objects as adeptly as humans. Wei’s major contributions include the development of GelStereo, a high-resolution visuotactile sensor that uses stereo vision for in-hand object localization, and novel deep learning architectures for visual-tactile fusion—such as self-attention and 3D convolution networks—that predict grasp outcomes and assess grasp states for deformable objects. His research on slip detection using a Generalized Visual-Tactile Transformer and on intelligent roughness and slip recognition via patterned tactile sensors has advanced robotic grasping reliability. With over 300 citations across his most-cited works, Wei’s impact is evident. Notable achievements include his work on meta-residual policy learning for zero-trial skill adaptation and bridging vision-language models with real-world long-horizon manipulations. His contributions are pivotal for developing autonomous robots capable of delicate, adaptive manipulation in unstructured environments.
Research Focus
Key Achievements
Top Papers
- 1In-Hand Object Localization Using a Novel High-Resolution Visuotactile Sensor103 citations · 2021
- 2
- 3
- 4
- 5
- 6
- 7Generalized Visual-Tactile Transformer Network for Slip Detection15 citations · 2020
- 8
- 9SOZIL: Self-Optimal Zero-Shot Imitation Learning2 citations · 2021
- 10