Yunfeng Lin

Papers

Total Citations

H-Index

About

Yunfeng Lin is a leading researcher in embodied AI and robotic perception, with a focus on enabling legged robots to navigate complex, unstructured environments. His major contributions lie at the intersection of world models, visual perception, and scalable simulation for robot learning. In his highly cited work, "World Model-Based Perception for Visual Legged Locomotion," Lin tackles the challenge of data-inefficient learning from high-dimensional visual inputs, proposing a framework that integrates proprioception and vision for robust terrain adaptation. This work has garnered 5 citations since 2025, reflecting its immediate impact on the field. More recently, Lin introduced "GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMs," which addresses the critical bottleneck of human effort in creating diverse simulation tasks. By leveraging large language models for multi-modal reasoning, GenSim2 enables scalable, multi-task policy training, bridging the sim-to-real gap. With only 2 citations since 2024, this work is poised to become a cornerstone for efficient robotic simulation. Lin’s research is notable for its practical focus on overcoming real-world deployment hurdles, making him a rising figure in robotics and embodied intelligence.