答案已发布4小时前Last edited 3小时前16 来源

What is the Westlake World Model Research Center launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, as reported by IT Home citing the university, and how do its university level interdisciplinary role, director Yu Kaicheng, chief scientist Jin Yaochu

The Westlake World Model Research Center is a university level interdisciplinary research platform launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, focused on world models and physical intelligence—not simply Its stated aim is to help AI move from understanding language toward understa...

发布者图片由 GPT Image 2 生成

使用 Studio Global AI 搜索并核查事实浏览更多热门页面

研究答案

要点

The Westlake World Model Research Center is a university level interdisciplinary research platform launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, focused on world models and physical intelligence—not simply
Its stated aim is to help AI move from understanding language toward understanding, predicting and acting in the physical world.
[8] Mission and leadership Organization: Yu Kaicheng (于开丞) is director, Jin Yaochu is chief scientist, and multiple principal investigators are to share responsibility for different research directions; the university wide platform is inten
[8] Leadership backgrounds: Yu is a Westlake School of Engineering assistant professor and doctoral supervisor, and his university biography also identifies him as KMind’s co founder and chief scientist; Jin is identified in the launch repo

What is the Westlake World Model Research Center launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, as reportedAI-generated editorial hero image for What is the Westlake World Model Research Center launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, as reported.
AI 提示
Create a landscape editorial hero image for this Studio Global article: What is the Westlake World Model Research Center launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, as reported. Article summary: The Westlake World Model Research Center is a university level interdisciplinary research platform launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, focused on world models and physical inte. Topic tags: general web, ai, automation, api, growth. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts wit
openai.com

The Westlake World Model Research Center is a university-level interdisciplinary research platform launched at Westlake University’s Yungu Campus in Hangzhou on October 31, 2026, focused on world models and physical intelligence—not simply another language-model laboratory. Its stated aim is to help AI move from understanding language toward understanding, predicting and acting in the physical world. 8

Mission and leadership

Organization: Yu Kaicheng (于开丞) is director, Jin Yaochu is chief scientist, and multiple principal investigators are to share responsibility for different research directions; the university-wide platform is intended to bring together research across disciplines and connect basic theory, key technologies and applications. 8
Leadership backgrounds: Yu is a Westlake School of Engineering assistant professor and doctoral supervisor, and his university biography also identifies him as KMind’s co-founder and chief scientist; Jin is identified in the launch report as Westlake’s chair professor of artificial intelligence and head of its AI department. 3 8
Core bottlenecks: The center targets intelligent systems’ difficulties in understanding physical laws, predicting environmental changes and making decisions through interaction, with the goal of enabling understanding of the world, prediction of future states and autonomous action in real environments. 8
Integrated approach: Its announced research framework combines perception, cognition, prediction and action rather than treating these as isolated capabilities. 8 In practical terms, the proposed concept-world-model approach represents physical concepts in an abstract mathematical latent space, allowing complex scenes to be decomposed and recombined to reason about unfamiliar situations rather than merely generate their pixels. 15

How Yu’s earlier work connects

Education and industry experience: Yu earned a first-class-honours engineering bachelor’s degree at the University of Hong Kong in 2016 and a doctorate in computer and communication sciences at EPFL in 2021; his university biography lists experience at Intel Intelligent Systems Lab, Abacus.AI and Alibaba’s DAMO Academy, alongside research in 3D vision, autonomous driving and automated machine learning. 3
AutoLab foundation: After leaving Alibaba and joining Westlake in 2023, he established the Autonomous Intelligence Laboratory, AutoLab, and began world-model research; the initial route generated extreme-scenario videos to train autonomous-driving models, and reporting credits the team with winning an ECCV autonomous-driving competition in 2024. 15
Late-2024 change: AutoLab then shifted toward “concept world models,” replacing pixel generation with representations of physical concepts in latent space—a research direction directly aligned with the new center’s physical-intelligence focus. 15 58
Commercial connection: Reporting links that concept-world-model work to the startup Awomo, indicating a route from university research toward industrial deployment. 58 The available excerpts support formation in early 2026, but provide insufficient evidence to verify the specific January date. 7

What remains uncertain or unannounced

The IT Home launch account names Yu and Jin but does not identify the other PIs or provide a detailed initial project portfolio. 8
That does not establish that no initial outputs were announced elsewhere: a separate report attributed to People’s Daily describes a first world-model version, a pipeline processing more than 500,000 hours of embodied physical-world data, and a geographically distributed computing platform approaching 1,000 accelerators. 5
Insufficient evidence in the supplied excerpts verifies Yu’s precise authorship and citation record for BEVFusion, or a detailed implementation for predicting action outcomes and continually learning from real-world feedback. These should be distinguished from the center’s announced research ambitions.

Studio Global AI

继续你的研究

此页面包含一个有来源支持的答案，您可以在 Studio Global 内继续。

使用 Studio Global AI 搜索并核查事实

人们还问