Zheyuan Hu
Papers
4
Total Citations
23
H-Index
2
About
Zheyuan Hu is a roboticist advancing the frontier of dexterous manipulation, particularly for multi-fingered hands operating in complex, contact-rich environments. His work bridges reinforcement learning and computer vision to tackle some of robotics' hardest problems. His most impactful contribution, “Dexterous Manipulation from Images: Autonomous Real-World RL via Substep Guidance” (2023, 13 citations), introduces a novel framework that uses substep guidance to enable real-world reinforcement learning directly from visual input, bypassing the need for precise simulations or engineered rewards. This approach has been pivotal for tasks like underactuated object manipulation. Hu also developed “Object Pose Estimation for Robotic Grasping based on Multi-view Keypoint Detection” (2021, 6 citations), which addresses the persistent challenge of grasping cluttered and occluded objects by leveraging multi-view geometry for robust pose estimation. Earlier work on resource management for real-time edge systems (2020, 2 citations) and his 2023 paper “REBOOT: Reuse Data for Bootstrapping Efficient Real-World Dexterous Manipulation” (2 citations) further demonstrate his commitment to making robotic learning more sample-efficient and deployable. Hu’s research is essential reading for anyone interested in data-driven, real-world dexterous manipulation.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3A Resource Management Model for Real-time Edge System of Multiple Robots2 citations · 2020
- 4