Stanford Hosts Physical AI & Robotics Event

PL-Universe's Founder & COO, Ge Jin, presented a new approach for intelligent manufacturing: a "universal ontology + rapidly replaceable dedicated end-effectors" solution. This system is designed to meet the high industrial demands for flexibility and reliability in large-scale deployments. Quan Kuichen, head of the company's Large Model Team, detailed their technical progress in applying Vision-Language-Action (VLA) models to industry. Key breakthroughs were shared in areas like multi-modal data collection, cloud-edge collaboration, and few-shot learning to achieve sub-millimeter precision. The event also addressed the global landscape of Physical AI, with automotive industry observer Xing Lei noting that China's strength lies in supply chains and application scenarios. In contrast, the US currently leads in the development of algorithms and chips for the embodied AI field. From a venture capital standpoint, TSVC General Partner Spencer Greene offered insights on investing in the sector. He pointed to structural labor shortages as a key driver for Embodied AI systems and stressed the importance of focusing on real commercial value over the hype surrounding humanoid robots.