VLA-Precision: Asymmetric Co-Bootstrapping for Efficient Real-World Online RL of Vision-Language-Action Models

Researchers introduced VLA-Precision, a method using asymmetric co-bootstrapping for efficient real-world online reinforcement learning of vision-language-action models, addressing the unreliability of pretrained VLA models in tasks requiring precision and repeatability. The approach enables autonomous trial-and-error improvement of VLA post-training beyond demonstrations alone.

Pretrained vision-language-action VLA models enable broad manipulation but remain unreliable in tasks demanding precision and repeatability. Applying real-world online reinforcement learning RL to VLA post-training enables autonomous trial-and-error improvement beyond demonstrations alone, but e