Submitted by Polina Fedotova 191 Green-VLA: Staged Vision-Language-Action Model for Generalist Robots Sber Robotics Center 29 3