Submitted by Polina Fedotova 216 Green-VLA: Staged Vision-Language-Action Model for Generalist Robots Sber Robotics Center 31 4