Yunfan Jiang
Yunfan Jiang is a PhD student in Computer Science at Stanford University, advised by Professor Fei-Fei Li, and a graduate research assistant in the Stanford Vision and Learning Lab. He is a recipient of the NVIDIA Graduate Fellowship for 2025 and has served as a research intern at NVIDIA, where he contributed to the GR00T team developing robot foundation models. He is best known as lead author of VIMA, a multimodal prompt-based framework for general robot manipulation, and for his research on sim-to-real transfer and whole-body manipulation for everyday household activities.
“RoboTTT pretrained at 8K timesteps achieves 71.5% average task completion score, which is 63% higher than the same model pretrained at 1K timesteps (43.9%) and 57% higher than the best short-context baseline (45.6%)”
Source→“Lead author, previously first author on VIMA (multimodal prompt-based robot manipulation). His trajectory from prompt-conditioned manipulation to long-context TTT policies represents a coherent research program”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.