VLA ModelsarXiv preprint2025
RLinf-VLA: A Unified and Efficient Framework for Reinforcement Learning of Vision-Language-Action Models
Elena Sorina Lupu, Patrick Spieler, Khurram Javed, Kris De Asis, John D. Martin, Martha Steenstrup, Joseph Modayil et al.Stanford University
摘要
We present RLinf-VLA, a unified framework for applying reinforcement learning to Vision-Language-Action models. Our approach enables efficient RL fine-tuning of VLAs through a novel reward shaping and policy optimization scheme, significantly improving robot manipulation performance with minimal additional data.
reinforcement learningVLAunified frameworkefficient trainingvision-language-action
