Ziye Hu

About

Ziye Hu is a pioneering robotics researcher whose work focuses on enabling robots to learn new skills rapidly from minimal human demonstrations—a capability inspired by human and animal learning. His core research areas span meta-learning, few-shot imitation learning, and domain-adaptive robotic manipulation. Hu’s major contributions include developing model-agnostic meta-learning frameworks enhanced with noise mechanisms for one-shot imitation (10 citations), and creating few-shot instance grasping systems that allow robots to identify and grasp specific novel objects in cluttered environments (9 citations). He has also advanced domain-adaptive meta-learning, introducing methods like Random Domain-Adaptive Meta-Learning (8 citations) and Replayed Task-Contrastive Meta-Learning (5 citations) to help robots transfer skills across different visual environments. His work addresses the fundamental challenge of making robots as adaptable as humans, with particular emphasis on learning from visual demonstrations across varying domains. Hu’s research is notable for its practical approach to the “learning to learn” paradigm, directly tackling real-world robotics problems where data is scarce and environments are unpredictable.

Research Focus

Key Achievements

Top Papers

1. 1
   Two-Stage Model-Agnostic Meta-Learning With Noise Mechanism for One-Shot Imitation

   10 citations · 2020
2. 2
   Few-Shot Instance Grasping of Novel Objects in Clutter

   9 citations · 2022
3. 3
   Learning With Dual Demonstration Domains: Random Domain-Adaptive Meta-Learning

   8 citations · 2022
4. 4
   Learning From Visual Demonstrations via Replayed Task-Contrastive Model-Agnostic Meta-Learning

   5 citations · 2022
5. 5
   Learning From Demonstrations Via Multi-Level and Multi-Attention Domain-Adaptive Meta-Learning

   4 citations · 2022

Key Collaborators

Contact & Links