Yanjiang Guo
Papers
3
Total Citations
21
H-Index
3
About
Yanjiang Guo is a researcher at the forefront of embodied AI and robot learning, with a focus on bridging the gap between large-scale pretrained models and real-world robotic control. His work centers on three key areas: vision-language-action (VLA) models, meta-reinforcement learning for policy transfer, and grounding language models in physical execution. In his 2025 study on improving VLA models with online reinforcement learning (9 citations), he demonstrated how to enhance large vision-language models beyond supervised fine-tuning, enabling more adaptive robotic manipulation. His 2023 paper on zero-shot policy transfer (8 citations) introduced a novel approach using disentangled task representations, allowing robots to compose prior experiences and generalize to unseen tasks—a crucial step toward human-like learning efficiency. Additionally, his work on DoReMi (4 citations) tackles the critical problem of plan-execution misalignment in LLM-guided robotics, detecting and recovering from failures during physical execution. Guo’s contributions are shaping how robots can leverage foundation models while maintaining robustness in the real world, making his research highly relevant for students and researchers working at the intersection of language, vision, and action.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3