VLA-Precision: Asymmetric Co-Bootstrapping for Efficient Real-World Online RL of Vision-Language-Action Models Paper • 2609.04355 • Published 11 days ago • 2