MotionVLA: Vision-Language-Action Model for Humanoid Motion Paper • 2606.15142 • Published Jun 13 • 5