Gemini Robotics
Gemini Robotics是由Google DeepMind[1]與Apptronik合作開發的先進「視覺-語言-行動」模型。[2]該模型以Gemini 2.0大型语言模型為基礎。[3]專門針對機器人應用進行調整，能理解並應對新的情境[4][5]另有一個相關版本名為Gemini Robotics-ER，代表「具身推理」。[3]兩款模型於2025年3月12日正式發佈[5]
2025年6月24日，Google DeepMind推出Gemini Robotics On-Device，這是專為在機器人裝置上本地運行而設計並優化的版本。[6]
目前，Gemini Robotics模型僅向特定受信任的測試者開放，包括Agile Robots、Agility Robots、Boston Dynamics與Enchanted Tools。[2]
參考資料
[编辑]- ^ Gemini Robotics. deepmind.google. [2025-03-12].
- ^ 2.0 2.1 Parada, Carolina. Gemini Robotics brings AI into the physical world. Google DeepMind. [2025-07-11].
- ^ 3.0 3.1 Knight, Will. Google's Gemini Robotics AI Model Reaches Into the Physical World. WIRED. May 12, 2025 [2025-03-12].
- ^ Google introduces new AI models for rapidly growing robotics industry. Reuters. March 12, 2025 [2025-03-12].
- ^ 5.0 5.1 Roth, Emma. Google DeepMind's new AI models help robots perform physical tasks, even without training. The Verge. March 12, 2025 [2025-03-12].
- ^ Parada, Carolina. Gemini Robotics On-Device brings AI to local robotic devices. Google DeepMind. [11 July 2025].