Galaxy General Unveils GraspVLA: The Premier Comprehensive End-to-End Embodied Grasping Foundational Large Model
2025-01-10

Galaxy General, in collaboration with the Beijing Academy of Artificial Intelligence, Peking University, and the University of Hong Kong, has introduced GraspVLA, a groundbreaking model that stands as the world's first comprehensive and generalized end-to-end embodied grasping foundational large model. This model seamlessly integrates perception, learning, and environment interaction capabilities. Leveraging a pre-training and post-training architecture, GraspVLA is pre-trained on billions of frames of data, empowering it to excel in zero-shot testing and demonstrate remarkable generalization capabilities. Officials have further established seven "gold standards" for generalization, encompassing illumination generalization, background generalization, planar position generalization, spatial height generalization, action strategy generalization, dynamic interference generalization, and object category generalization. These standards ensure that the model performs exceptionally well in a diverse range of environments.