MAIN-VLA: Modeling Abstraction of Intention and eNvironment for Vision-Language-Action Models Paper • 2602.02212 • Published Feb 2 • 1