「万物抓取」Demo
Search documents
万物皆可抓取?星海图G0 Plus让机器人“即看即会”,零样本时代来了
机器人大讲堂· 2026-01-07 00:00
Core Insights - The article discusses the breakthrough of the G0 Plus model by Xinghai Tu, which allows robots to understand and execute natural language commands for tasks like picking up objects, marking a significant advancement in embodied intelligence [1][3][5]. Group 1: Technological Breakthroughs - The G0 Plus model demonstrates zero-shot capabilities, enabling universal grasping without task-specific training, thus addressing the gap between advanced robotic models and practical applications [3][5]. - The model's ability to understand concepts like "grabbability," "stability," and "spatial relationships" enhances its operational intelligence, allowing it to perform tasks in real-world environments [7][10]. Group 2: Productization and Market Impact - The VLA all-in-one machine solution simplifies deployment for developers, allowing them to experience the G0 Plus capabilities within 30 minutes of unboxing, thus lowering the technical barrier for adoption [10][12]. - Collaborations with industry leaders like Huawei and BYD aim to validate the technology's reliability and facilitate large-scale applications in sectors such as industrial manufacturing and logistics [12][14]. Group 3: Strategic Vision - Xinghai Tu's strategy emphasizes vertical integration from motor control to AI model development, ensuring maximum system performance and user experience [18][20]. - The release of G0 Plus and the VLA all-in-one machine represents a pivotal moment in embodied intelligence, potentially leading to widespread adoption and a shift in market dynamics towards more user-friendly solutions [21][23].