暫無精選機器人影片
暫無影片
該公司尚未關聯精選機器人影片。
暫無主要機器人方向資訊。
Google DeepMind has launched the Gemini Robotics 2 family, a suite of three foundation models designed to bring whole-body control and advanced reasoning to humanoid robots. The release includes a vision-language-action (VLA) model capable of controlling 22-degree-of-freedom multifingered hands, an embodied reasoning (ER) model for multi-robot coordination, and an on-device variant that adapts to new hardware in hours. Demonstrated on platforms from Apptronik, Franka, and others, the models show significant progress in intricate manipulation tasks, though published benchmarks reveal the persistent difficulty of fine motor skills like screwing in a light bulb. The ER 2 model is now available in public preview via Google AI Studio, signaling a shift toward broader developer access in the robotics foundation model space.
該公司已關聯機器人,但仍在審核中,因此暫不對外顯示。論壇、圖片和影片會在機器人發布後出現。
領導層資訊暫不可用。