Yuying Ge

Papers

Total Citations

H-Index

About

Yuying Ge is a researcher advancing the frontier of generalizable robot learning, with a focus on integrating 3D scene understanding and vision-language models into robotic manipulation. Her work addresses the long-standing challenge of enabling robots to perform diverse tasks in unstructured real-world environments. In her highly cited paper "GNFactor: Multi-Task Real Robot Learning with Generalizable Neural Feature Fields" (2023, 11 citations), Ge pioneered a method that combines neural feature fields with multi-task learning, allowing robots to grasp both the 3D structure and semantic meaning of scenes—a critical step toward adaptable, real-world autonomy. Her follow-up work, "Policy Adaptation from Foundation Model Feedback" (2023, 4 citations), leverages pre-trained vision-language models to enable instruction-conditioned policies that generalize across different objects without task-specific retraining. By bridging foundation models with policy learning, Ge’s research demonstrates how large-scale pre-trained knowledge can be effectively transferred to physical robots, significantly reducing the need for extensive in-domain data. Her contributions are shaping a new paradigm where robots can understand and act in open-world environments, making her a notable emerging voice in the intersection of computer vision, robotics, and foundation model research.