--- 格式版本: 2 标题: "NVIDIA GTC 2026 物理 AI 精选会议" 原文链接: "https://www.nvidia.cn/gtc/sessions/physical-ai-days/" 发布日期: "2026-09-04" 发布时间校准状态: "found" 发布时间需复核: "否" 发布时间来源: "rule:local:strict_original_body" 发布时间证据: "Published Time: Fri, 04 Sep 2026 08:14:04 GMT" 发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body" 发布时间校准置信度: "high" 发布时间候选数量: 4 发布时间严格候选数量: 1 发布时间原页读取状态: "source template page reused from URL open" 发布时间未找到原因: "" 发布时间校准时间: "2026-09-06T01:05:02+08:00" 发布时间仲裁状态: "skipped" 发布时间仲裁尝试次数: 0 发布时间仲裁耗时毫秒: 0 发现时间: "2026-09-06T00:24:20+08:00" 入库时间: "2026-09-05T17:10:42.675Z" 来源平台: "固定入口" 搜索渠道: "fixed_url" 搜索词: "https://www.nvidia.cn/gtc/" 匹配关键词: - "AI" - "部署" - "deployment" - "性能" - "performance" - "GPU" 相关厂家: - "NVIDIA" - "字节" - "Hon Hai" - "Foxconn" 相关专家: [] 内容类型: "网页" 抓取工具: "Jina Reader" 清洗工具: "Jina Reader Markdown + Defuddle/Readability 正文提取" 原始附件: [] AI优质: "否" AI打分: 38 AI分档: "非优质" AI质检状态: "不通过" AI打分理由: "正文为NVIDIA GTC 2026物理AI精选会议议程页和网站导航，主线是数字孪生、机器人、汽车等物理AI应用会议，不讨论超节点/AI Rack/机架级基础设施；除Foxconn AI工厂和Samsung 50,000 GPU等零散提及外无rack-scale架构、互连、供电、液冷或部署规格。来源为NVIDIA官方会议页面但页面仅为议程摘要，无新品发布、标准、实测、量产或路线图新增事实，历史知识库也未显示其贡献。命中强否决项：应用/会议综述且无可外推机架级基础设施机制，实质为摘要壳和导航页，故非优质。" AI质检模型: "zj-deepseek-v4-flash" AI质检时间: "2026-09-06T01:10:52+08:00" AI主题相关性: 5 AI来源权威性: 12 AI新颖性: 5 AI技术细节: 5 AI商业部署信号: 5 AI完整性: 6 AI评分提示词版本: "v17-精简生产版" AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d" AI评分知识库版本: "knowledge_base_v1-20260819+runtime.94" AI评分知识库SHA256: "e0367b42beb690d54c0aa046486e5849e6a9ccabdeb59e7e9718079f0c03d394" AI评分知识库检索词: "[\"NVIDIA\",\"NVIDIA GTC\",\"https://www.nvidia.cn/gtc/\",\"RAS\",\"NPU\",\"GPU\",\"英伟达\",\"Intel\",\"GTC\",\"URL\",\"GMT\",\"EN\"]" AI评分知识库命中: "[{\"id\":\"historical-jun-025\",\"title\":\"英伟达、谷歌与国产超节点的三种网络选择\",\"sourceType\":\"curated_item\",\"time\":\"2026-06\",\"matchedTerms\":[\"NPU\",\"GPU\",\"英伟达\"],\"rank\":-13.64702724022332},{\"id\":\"historical-jan-apr-01\",\"title\":\"一、NVIDIA GTC 2026 相关基础设施发布与展示\",\"sourceType\":\"curated_item\",\"time\":\"2026-01_to_2026-04\",\"matchedTerms\":[\"NVIDIA\",\"NVIDIA GTC\",\"GPU\",\"GTC\",\"EN\"],\"rank\":-12.758737814354888},{\"id\":\"runtime-8870e60555ebf4805990c527\",\"title\":\"How XPUs Meet a World-Class AI Factory\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"NVIDIA GTC\",\"RAS\",\"GPU\",\"Intel\",\"GTC\",\"URL\",\"EN\"],\"rank\":-12.658347853117544},{\"id\":\"runtime-f1ae4283a0a763e97d550f0d\",\"title\":\"As AI Increases Demands on Memory, Storage Steps Up\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"NVIDIA GTC\",\"RAS\",\"GPU\",\"Intel\",\"GTC\",\"URL\",\"EN\"],\"rank\":-12.4854797041621},{\"id\":\"runtime-60eee3fc9750e1dd241e246e\",\"title\":\"Why Scaling AI Compute Performance Requires a New Power Architecture\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"NVIDIA\",\"NVIDIA GTC\",\"RAS\",\"GPU\",\"GTC\",\"URL\",\"EN\"],\"rank\":-12.056943769527518}]" 采集批次: "2026年9月6日0点59分28秒" 采集批次ID: "20260906-005928-61f855f3" 去重键: "https://www.nvidia.cn/gtc/sessions/physical-ai-days" --- Title: NVIDIA GTC 2026 物理 AI 精选会议 URL Source: https://www.nvidia.cn/gtc/sessions/physical-ai-days/ Published Time: Fri, 04 Sep 2026 08:14:04 GMT Markdown Content: Visit your regional NVIDIA website for local content, pricing, and where to buy partners specific to your country. [Continue](https://www.nvidia.com/) 即将举行：探索其他 GTC 大会：[**GTC 柏林** 10 月 20 — 22 日](https://www.nvidia.com/en-eu/gtc/) | [**GTC 华盛顿特区** 11 月 30 日 — 12 月 3 日](https://www.nvidia.com/gtc/dc/) | [**GTC 2027** 3 月 15 — 18 日](https://www.nvidia.cn/gtc/) * [](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [](https://www.nvidia.cn/account/) * [登录](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#)[LogOut](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [简中](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [EN](https://www.nvidia.com/gtc/) * [한국어](https://www.nvidia.com/ko-kr/gtc/) * [日本語](https://www.nvidia.com/ja-jp/gtc/) * [繁中](https://www.nvidia.com/zh-tw/gtc/) * [简中](https://www.nvidia.cn/gtc/) * [DE](https://www.nvidia.com/de-de/gtc/) * [FR](https://www.nvidia.com/fr-fr/gtc/) 平台 其他链接 [](https://www.nvidia.cn/gtc/) [主题演讲](https://www.nvidia.cn/gtc/keynote/) 探索 * [大会主题](https://www.nvidia.cn/gtc/conference-topics/) * [海报展示](https://www.nvidia.cn/gtc/posters/) * [演讲嘉宾](https://www.nvidia.cn/gtc/speakers/) * [创业公司和投资机构](https://www.nvidia.cn/gtc/startups/) * [培训和认证](https://www.nvidia.cn/gtc/training/) [赞助商和参展商](https://www.nvidia.cn/gtc/sponsors/) 更多 * [行为准则](https://www.nvidia.cn/gtc/code-of-conduct/) * [联系我们](https://www.nvidia.cn/gtc/contact/) * [常见问题](https://www.nvidia.cn/gtc/faq/) * [隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/) * [查看所有 GTC 活动](https://www.nvidia.cn/gtc/events/) * [主题演讲](https://www.nvidia.cn/gtc/keynote/) * 探索 * [大会主题](https://www.nvidia.cn/gtc/conference-topics/) * [海报展示](https://www.nvidia.cn/gtc/posters/) * [演讲嘉宾](https://www.nvidia.cn/gtc/speakers/) * [创业公司和投资机构](https://www.nvidia.cn/gtc/startups/) * [培训和认证](https://www.nvidia.cn/gtc/training/) * [赞助商和参展商](https://www.nvidia.cn/gtc/sponsors/) * 更多 * [行为准则](https://www.nvidia.cn/gtc/code-of-conduct/) * [联系我们](https://www.nvidia.cn/gtc/contact/) * [常见问题](https://www.nvidia.cn/gtc/faq/) * [隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/) * [查看所有 GTC 活动](https://www.nvidia.cn/gtc/events/) * [](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [简中](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [EN](https://www.nvidia.com/gtc/) * [한국어](https://www.nvidia.com/ko-kr/gtc/) * [日本語](https://www.nvidia.com/ja-jp/gtc/) * [繁中](https://www.nvidia.com/zh-tw/gtc/) * [简中](https://www.nvidia.cn/gtc/) * [DE](https://www.nvidia.com/de-de/gtc/) * [FR](https://www.nvidia.com/fr-fr/gtc/) [点播观看](https://www.nvidia.cn/on-demand/search/?facet.event_name[]=GTC%20San%20Jose&facet.event_year[]=2026&facet.mimetype[]=event%20session&headerText=All%20Sessions&layout=list&page=1&q=-&sort=relevance&sortDir=desc>cnavinherit=true)[](https://www.nvidia.cn/gtc/sessions/physical-ai-days/# "Menu") [点播观看](https://www.nvidia.cn/on-demand/search/?facet.event_name[]=GTC%20San%20Jose&facet.event_year[]=2026&facet.mimetype[]=event%20session&headerText=All%20Sessions&layout=list&page=1&q=-&sort=relevance&sortDir=desc>cnavinherit=true)[](https://www.nvidia.cn/gtc/sessions/physical-ai-days/# "Menu") * [主题演讲](https://www.nvidia.cn/gtc/keynote/) * [探索](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [探索](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [大会主题](https://www.nvidia.cn/gtc/conference-topics/) * [海报展示](https://www.nvidia.cn/gtc/posters/) * [演讲嘉宾](https://www.nvidia.cn/gtc/speakers/) * [创业公司和投资机构](https://www.nvidia.cn/gtc/startups/) * [培训和认证](https://www.nvidia.cn/gtc/training/) * [赞助商和参展商](https://www.nvidia.cn/gtc/sponsors/) * [更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/#) * [行为准则](https://www.nvidia.cn/gtc/code-of-conduct/) * [联系我们](https://www.nvidia.cn/gtc/contact/) * [常见问题](https://www.nvidia.cn/gtc/faq/) * [隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/) * [查看所有 GTC 活动](https://www.nvidia.cn/gtc/events/) ![Image 1](https://www.nvidia.cn/content/dam/targeted-agenda/hero-banner/gtc26-target-agenda-multi-img-md460-d.jpg) # 物理 AI 精选会议 探索为期两天的精心策划的会议，来自各行业的专家以及开发者通过多场技术与应用并重的主题演讲和案例分享， 展示工业 AI、机器人技术、汽车和数字孪生如何推动物理 AI 的变革，帮助您把握 AI 与真实物理世界深度融合的发展方向与行业机遇。 [查看更多物理 AI 精选会议](https://www.nvidia.cn/gtc/session-catalog/?topics=Simulation%20%2F%20Modeling%20%2F%20Design%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%203D%20Data%20Interoperability%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20BioPharma%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20CAE%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Climate%20%2F%20Weather%20%2F%20Ocean%20Modeling%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Computational%20Chemistry%20%2F%20Materials%20Science%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Industrial%20Digitalization%20%2F%20Digital%20Twin%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Molecular%20Dynamics%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Physics%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Quantum%20Computing%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Supercomputing%2CSimulation%20%2F%20Modeling%20%2F%20Design%20-%20Synthetic%20Data%20Generation&technicalLevels=General%20Interest®code=no-ncid&ncid=no-ncid) ![Image 2](https://www.nvidia.cn/content/nvidiaGDC/cn/zh_CN/gtc/sessions/physical-ai-days/_jcr_content/root/responsivegrid/nv_container/nv_container_temp3/nv_image.coreimg.100.410.jpeg/1769040812858/gtc-dc25-physical-ai-and-robotics-industrial-ai-kitchen-1080.jpeg) ![Image 3](https://www.nvidia.cn/content/nvidiaGDC/cn/zh_CN/gtc/sessions/physical-ai-days/_jcr_content/root/responsivegrid/nv_container/nv_container_temp3/nv_image_one.coreimg.100.190.png/1771924385209/gtc26-simulation-modeling-design-apollo-factory-auto-ignite.png) ![Image 4](https://www.nvidia.cn/content/nvidiaGDC/cn/zh_CN/gtc/sessions/physical-ai-days/_jcr_content/root/responsivegrid/nv_container/nv_container_temp3/nv_image_two.coreimg.100.190.png/1769783246845/gtc26-auto-safety-session-1080x1080.png) ## 精选会议 Talk On-Demand ## **Accelerate the Physical AI Era With Digital Twins and Real-Time Simulation [S81598]** **Rev Lebaredian** | VP of Omniverse and Simulation Technology | NVIDIA Join Rev Lebaredian, NVIDIA’s VP of Omniverse and simulation technology, as he shares how autonomous, OpenClaw-based agents, digital twins and real-time simulation—powered by OpenUSD and NVIDIA’s accelerated computing platform, libraries, and ecosystem—are accelerating breakthroughs across industries in the era of physical AI. Discover their impact on everything from product design and manufacturing to robotics, autonomous vehicles, smart cities, climate science, and space exploration. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764890417617001sMfV)**Tuesday Mar 17 | 5:00 PM - 5:50 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Panel On-Demand ## **Building the Future of Manufacturing [S81573]** **Rev Lebaredian** | VP of Omniverse and Simulation Technology | NVIDIA **Seungyong Doh** | EVP, Head of Digital Transformation | SK hynix **Zhaopeng Chen** | CEO and Founder | Agile Robots **Rainer Brehm** | COO of Automation Business and CTO of Digital Industries | Siemens **Marc Segura** | President | ABB Robotics **Roslyn Stuart** | VP of Synergy Creation | Hitachi Digital The manufacturing industry is entering a new era, driven by AI, robotics, and simulation. Join distinguished leaders from the semiconductor and electronics, industrial and process manufacturing, and robotics industries as they share their pioneering use cases and how these technologies enable them to innovate, reimagine production systems, and transform the design and operation of advanced manufacturing and warehouse facilities. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764814786012001rwOF)**Tuesday Mar 17 | 6:00 PM - 7:00 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **How Open World Models are Powering the Next Breakthroughs in Physical AI [S81667]** **Ming-Yu Liu** | VP Generative AI Research | NVIDIA Generating the photorealistic data and robust reasoning capabilities needed for machines to understand real-world complexity remains one of the biggest bottlenecks in physical AI. NVIDIA Cosmos world foundation models are at the core of AI models and agents that can perceive, understand, and adapt to the real-world like never before. Learn how developers can use Cosmos open models and tools to accelerate training, customize pipelines, and deploy faster across robotics, autonomous vehicles, and vision AI agents. This session will also explore how the latest Cosmos updates advance physics-aware learning, unlock simulation-to-real capabilities, and enable long-running agents and frameworks like OpenClaw to orchestrate continuous synthetic data generation and reasoning workflows at scale. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764982322522001MHxP)**Tuesday Mar 17 | 10:00 PM - 10:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Building and Scaling AI Factories With Digital Twins and Robotics [S81508]** **Leo Guo** | GM | Hon Hai Technology Group (Foxconn) Foxconn is collaborating with NVIDIA to build an AI factory in Texas for the production of AI servers. Foxconn will leverage NVIDIA Omniverse libraries to build a digital twin environment for simulation and real-time monitoring to rapidly scale up production lines, and will introduce humanoid robots and a high degree of automation for the first time. All these will first be simulated in NVIDIA's environment and optimized in a visual setting to help Foxconn create a safer, more efficient, and more flexible manufacturing environment. It's committed to building this factory into a benchmark plant that can serve as a model for others to follow. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764474439539001PSZc)**Monday Mar 16 | 11:00 PM - 11:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Panel On-Demand ## **From Concept to Production: Humanoid Robotics at Scale [S81645]** **Amit Goel** | Head of Robotics and Edge Computing Ecosystem | NVIDIA **Ashok Elluswamy** | VP, AI Software | Tesla **Arnaud Robert** | CEO | Hexagon Robotics **Pras Velagapudi** | CTO | Agility **Deepak Pathak** | CEO and Co-Founder | Skild AI **Chelsea Finn** | Assistant Professor at Stanford University | Co-Founder at Physical Intelligence Humanoid robots are evolving from prototypes to production-ready systems, powered by breakthroughs in reasoning AI, large-scale simulation, and real-time edge computing. But to get there, the industry must bridge the gap between digital intelligence and physical reality. Hear from luminaries at the world’s leading robotics companies on the future of physical AI, and what it will take to unlock truly capable humanoids. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764965646517001oNAl)**Monday Mar 16 | 10:00 PM - 10:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **从视觉还原到物理交互： 物理可驱动3D 资产生成技术 Beyond Visuals: A Physics-Informed Generative Method for Articulated 3D Assets [S81935]** **Dongping Li** | Researcher | 字节跳动 本演讲将介绍一种将多模态输入转化为物理可驱动资产的生成式方法。我们将详细阐述如何利用 3D 大模型，从输入图像或自然语言，得到具有完整关节结构的3D资产，涵盖自动化关节估计、部件生成以及物理属性推断等技术。 This session introduces a generative method that transforms multimodal inputs into physically drivable assets. We will detail how we leverage 3D large models to transform images or natural language descriptions into fully articulated assets, featuring automated semantic joint estimation, part-aware mesh generation, and precise physical property inference. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1765421588426001VR6a)**Wednesday Mar 18 | 2:00 AM - 2:50 AM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Digital Twins at Scale: The New Retail Growth Infrastructure [S82455]** **Morgan Mao** | Global CEO | SKAI Intelligence **Charles KIM** | Digital Director, Korea | LVMH Retail brands are under pressure to produce high-quality content at unprecedented speed and scale without increasing costs, and this session demonstrates how AI-powered digital twin technology built with NVIDIA Omniverse addresses that challenge. By transforming physical products into hyper-realistic, physics-accurate virtual assets, retailers can instantly deploy them across marketing, advertising, and e-commerce. Using robotic capture systems pre-trained with NVIDIA Isaac Sim and AI lighting agents trained with NVIDIA Kaolin, products are converted into production-grade digital twins with optimized illumination and material realism. Real-world case studies illustrate how workflows built with Omniverse enable mass content creation, synthetic data generation, and real-time rendering—replacing traditional photoshoots with scalable, AI-driven pipelines that deliver measurable improvements in speed, efficiency, and performance. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1770962477886001PhSt)**Tuesday Mar 17 | 12:00 AM - 12:40 AM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Panel On-Demand ## **How to Build Planetary-Scale AI Infrastructure [S81671]** **Chris Dolan** | Chief Data Center Officer | Crusoe Energy Systems, Inc. **Scott Wallace** | Director Data Center Engineering | NVIDIA **Catherine Kniker** | Chief Marketing and Sustainability Officer | PTC **Natasha Nelson** | CTO of Services and VP of EcoStruxure Power | Schneider Electric USA **Vivek Mishra** | Corporate VP | Cadence Design Systems, Inc. Gigawatt-scale AI facilities are pushing data center infrastructure beyond the limits of traditional design and operations workflows. Siloed approaches to buildings, power, cooling, and compute make it difficult to scale efficiently while meeting energy and sustainability requirements. This panel explores how simulation-based co-design and physically accurate digital twins are being used to plan and operate large-scale AI infrastructure. Industry leaders will discuss how these methods help evaluate tradeoffs earlier, improve efficiency, and better align facility, power, thermal, and compute decisions. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764983273243001rHVB)**Tuesday Mar 17 | 11:00 PM - 11:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Panel On-Demand ## **The Future of Industrial Autonomy [S81844]** **Deepu Talla** | Vice President of Robotics and Edge AI | NVIDIA **Rodolphe Katra** | Chief AI Officer | Medtronic **Suresh Venkatarayalu** | CTO | Honeywell **Mariella Guerricchio** | Chief Innovation and Technology Officer | Hitachi Rail **Brandon Hootman** | Vice President, AI Platforms & Technologies | Caterpillar **Rob Smith** | CEO | KION GROUP AG **Patrick Kelleher** | CEO | GXO This panel brings together industry leaders from Honeywell Industrial, KION Group, Caterpillar, Medtronic, Hitachi Rail and GXO to explore how advanced AI and edge computing are reshaping the future of industrial and healthcare systems. Together, they will discuss the major trends accelerating edge intelligence—from autonomous operations and multi-modal sensor fusion to real-time decision-making and digital twins—and share high-impact AI use cases driving measurable business growth and societal value. The conversation will highlight challenges in deploying scalable, safety-critical edge AI and reveal how these global innovators are architecting their next-generation solutions using NVIDIA’s accelerated computing and end-to-end edge platform. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1765256653028001f0cd)**Wednesday Mar 18 | 4:00 PM - 5:00 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Disney’s Olaf: From the Screen to Reality via Physical AI [S81492]** **Moritz Baecher** | Director, Research Lab Zurich | Disney Research Imagineering The translation of animated characters like Olaf, or the design of characters like our BDX Droids, requires balancing artistic fidelity with the constraints of the physical world. This talk explores how these robotic characters achieve that balance through the integration of modular mechatronics and deep reinforcement learning. Leveraging GPU-accelerated simulation to handle complex mechanisms, we train control policies that enable our characters to self-balance and self-regulate temperature. We further demonstrate progress in believable, autonomous navigation and human-robot interaction driven by onboard perception. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764198377597001RxTQ)**Tuesday Mar 17 | 4:00 PM - 4:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **How Virtual Twins are Shaping the Next Industrial Revolution [S81501]** **Florence Hu-Aubigny** | Executive VP R&D | Dassault Systèmes AI-powered virtual twins, science-validated industry world models, and virtual companions are shaping the next frontier for industries. Learn how Dassault Systèmes’ industry world models, agentic 3DEXPERIENCE platform, and virtual twin technologies—combined with NVIDIA AI infrastructure, open models, and libraries—are establishing new ways of working and transforming biology, materials science, engineering, and manufacturing. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764238975556001go7e)**Tuesday Mar 17 | 9:00 PM - 9:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Transforming Semiconductor Manufacturing with Agentic AI from Design and Engineering to Production [S81834]** **Yong Ho Song** | Corporate EVP, Head of AI Center(DS) | Samsung Electronics **Shankar Krishnamoorthy** | Chief Product Development Officer | Synopsys Join Samsung experts to explore how AI and digital twins are reshaping semiconductor manufacturing—from electronic design automation (EDA) and computational lithography to the design and operation of advanced fabrication plants. Learn how Samsung builds solutions with NVIDIA CUDA-X libraries, including cuLitho, to achieve up to 20x speedups in key manufacturing processes. Explore how they leverage NVIDIA Omniverse libraries to build large-scale industrial digital twins to conduct closed loop simulations and predict and optimize fab scheduling, maintenance, and yield. Discover how AI agents, built on NVIDIA Cosmos world foundation models, are driving chip, process, and system-wide optimization and innovation. Learn how Samsung’s AI Factory, powered by over 50,000 NVIDIA GPUs, makes all of this possible. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1765250031381001C5EV)**Tuesday Mar 17 | 5:00 PM - 5:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Visual AI Agents for Real-Time Video Understanding [S81836]** **Roopa Prabhu** | Sr. Director of Engineering | NVIDIA **Adam Ryason** | Product Manager, NVIDIA AI Blueprint for Video Search and Summarization (VSS) | NIVIDIA The next generation of visual AI agents starts with NVIDIA Blueprint for video search and summarization (VSS), which enables the creation of interactive analytical agents that reason through massive amounts of video, assess physical processes, and provide decision-making insights to streamline operations. The blueprint combines powerful vision language models (VLMs), advanced search capabilities with large language models, and Cosmos Reason VLM into a full suite of software for building, We've made it easier to customize and deploy at the edge, on-premises, or in the cloud. Be the first to hear about the new features of VSS to accelerate your AI agent development. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1765250364742001WogX)**Tuesday Mar 17 | 5:00 PM - 5:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Physical AI for the Real World: A Vision From NVIDIA Robotics Research [S81479]** **Yashraj Narang** | Sr. Robotics Research Manager | NVIDIA **Jim Fan** | Director and Distinguished Research Scientist | NVIDIA **Yuke Zhu** | Director and Distinguished Research Scientist | NVIDIA One of the hardest problems in physical AI is transforming robots from task-specific machines to general-purpose, intelligent collaborators. NVIDIA Research is pioneering a unified AI- and simulation-driven approach to solve this challenge, working from theoretical advances all the way to high-performance deployment. Join this session to learn about research breakthroughs in sim-to-real, real-world learning, model architecture, and model training, as well as key milestones on the path to generalist embodied intelligence for real-world applications. Gain perspectives about where robotics is headed, what technical hurdles remain, and how the next wave of research innovation could unlock entirely new markets. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764122135072001YB9P)**Tuesday Mar 17 | 8:00 PM - 8:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **From Research to Production: How Alpamayo Accelerates Autonomous Vehicle Development [S81779]** **Marco Pavone** | Sr. Research Director | NVIDIA **Patrick Liu** | Director, Autonomous Driving | NVIDIA Discover the latest advancements in end-to-end autonomous driving as the industry accelerates toward scalable, Level 4 AI-driven systems. This session explores how NVIDIA’s Alpamayo—particularly reasoning vision-language-action models (VLA)—enable vehicles to interpret, reason about, and navigate complex real-world scenarios with greater humanlike understanding. We’ll highlight how this approach is strengthened by a continuous, self-improving loop powered by NVIDIA Cosmos and agentic AI, and safeguarded by NVIDIA Halos, NVIDIA’s comprehensive full-stack safety system for physical AI. Finally, we’ll discuss how NVIDIA is building an open ecosystem to empower partners in developing next-generation, robotaxi-ready autonomous vehicles. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1765231458687001uO8N)**Monday Mar 16 | 10:00 PM - 10:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Talk On-Demand ## **Visual AI Agents for Real-Time Video Understanding [S81836]** **Roopa Prabhu** | Sr. Director of Engineering | NVIDIA **Adam Ryason** | Product Manager, NVIDIA AI Blueprint for Video Search and Summarization (VSS) | NIVIDIA The next generation of visual AI agents starts with NVIDIA Blueprint for video search and summarization (VSS), which enables the creation of interactive analytical agents that reason through massive amounts of video, assess physical processes, and provide decision-making insights to streamline operations. The blueprint combines powerful vision language models (VLMs), advanced search capabilities with large language models, and Cosmos Reason VLM into a full suite of software for building, We've made it easier to customize and deploy at the edge, on-premises, or in the cloud. Be the first to hear about the new features of VSS to accelerate your AI agent development. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1765250364742001WogX)**Tuesday Mar 17 | 5:00 PM - 5:40 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** Tutorial ## **An Introduction to Robot Simulation [S81488]** **Rishabh Chadha** | Technical Marketing Engineer - Isaac | NVIDIA **Asawaree Bhide** | Robotics Technical Marketing Engineer | NVIDIA Physical AI is transforming how robots interact with the world. Building these advanced systems requires three computers: compute for training, advanced simulation for validation, and efficient edge deployment. With a simulation-first approach, robots can train across millions of scenarios, generalize faster, and adapt to real-world complexity. In this technical deep-dive session, learn core robotics simulation workflows, from building virtual environments to configuring robots, sensors, and physics, using NVIDIA Isaac Sim and Isaac Lab, built with NVIDIA Omniverse libraries. You'll also explore how to generate and leverage synthetic data with NVIDIA Cosmos world foundation models to train and validate robot perception and control policies before deploying them in the real world. [Add to Schedule (opens in a new tab)](https://register.nvidia.cn/flow/nvidia/gtc26/apcn/page/catalogv/session/1764193070152001uLPh)**Thursday Mar 19 | 4:00 PM - 5:30 PM GMT** **[显示更多](https://www.nvidia.cn/gtc/sessions/physical-ai-days/)** 探索 * [大会主题](https://www.nvidia.cn/gtc/conference-topics/) * [海报展示](https://www.nvidia.cn/gtc/posters/) * [演讲嘉宾](https://www.nvidia.cn/gtc/speakers/) * [创业公司和投资机构](https://www.nvidia.cn/gtc/startups/) * [培训和认证](https://www.nvidia.cn/gtc/training/) 发现 * [主题演讲](https://www.nvidia.cn/gtc/keynote/) * [赞助商和参展商](https://www.nvidia.cn/gtc/sponsors/) 更多 * [行为准则](https://www.nvidia.cn/gtc/code-of-conduct/) * [联系我们](https://www.nvidia.cn/gtc/contact/) * [常见问题解答](https://www.nvidia.cn/gtc/faq/) * [隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/) * [查看所有 GTC 活动](https://www.nvidia.cn/gtc/events/) 关注 GTC [](https://space.bilibili.com/1320140761/)[](https://www.douyin.com/user/MS4wLjABAAAA-v_JG_5ioC7lMiXoo4bh1nCeK4TEvA1nJxrRZazrByQ)[](http://weibo.com/u/5943087025) ![Image 5](https://www.nvidia.cn/content/dam/en-zz/zh_cn/Solutions/newsroom/wc_qrcode.jpg?v2) 扫描二维码关注 NVIDIA 英伟达微信公众号 [](https://www.nvidia.cn/gtc/sessions/physical-ai-days/) CHN - 中国 * [隐私声明](https://www.nvidia.cn/about-nvidia/privacy-policy/) * [您的隐私选择](https://www.nvidia.cn/about-nvidia/privacy-center/) * [服务条款](https://www.nvidia.cn/about-nvidia/terms-of-service/) * [无障碍访问](https://www.nvidia.cn/about-nvidia/accessibility/) * [公司政策](https://www.nvidia.cn/about-nvidia/company-policies/) * [产品安全性](https://www.nvidia.com/en-us/security/) * [联系我们](https://www.nvidia.cn/contact/asia/) * [英文网站](https://www.nvidia.com/en-us/) Copyright © 2026 NVIDIA Corporation NVIDIA 及其第三方合作伙伴利用 Cookie 和其他工具收集并记录您提供的信息，以及您与我们网站的互动信息，以提高性能、进行分析并辅助营销工作。点击“全部接受”，即表示您同意我们使用 Cookie 和其他工具，如我们的 [Cookie 政策](https://www.nvidia.cn/about-nvidia/cookie-policy/)中所述。您可以通过点击“管理设置”来管理您的 Cookie 设置。请参阅我们的[隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/)，详细了解我们的隐私实践。 我们检测到全球隐私控制 (GPC) 信号，并记录了您拒绝本网站对此浏览器的所有可选 Cookie。您可以通过点击“管理设置”来管理您的 Cookie 设置。请参阅我们的 [Cookie 政策](https://www.nvidia.cn/about-nvidia/cookie-policy/) 以了解更多信息，同时查看问我们的[隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/)，以了解我们的隐私实践详情。 管理设置 拒绝可选 Cookie 全部接受 ![Image 7: 公司徽标](https://www.nvidia.cn/assets/cn/onetrust/consent/1984719c-04aa-48de-8c22-5174bdf5c0ed/01917434-8292-70f9-8fae-1602521d0b2a/logos/10ddf4ca-c072-45d0-b3ac-eead0ed93db0/b73a5c09-1e15-4dc7-8722-f9cb6c578110/9a7ca633-555d-4e05-a249-5cd7dc512da5/logo_nvidia.png) Cookie 设置 我们与第三方合作伙伴（包括社交媒体、广告和分析合作伙伴）使用 Cookie 和其他追踪技术，在您访问我们的网站时收集、存储、监控并处理您的某些相关信息。收集的信息可能与您、您的偏好或您的设备有关。我们使用这些信息来保障网站运行，并在征得您同意的情况下分析网站的性能和流量，以提供更加个性化的网络体验，并辅助我们的营销工作。点击下方不同类别的标题，获取更多信息，并根据您的偏好更改设置。您无法选择退出必备Cookie，因为其部署旨在确保我们网站的正常运行（例如提示 Cookie 横幅和记住您的设置等)。点击底部的“保存并接受”，即表示您同意按照您的设置使用我们的 [Cookie 政策](https://www.nvidia.cn/about-nvidia/cookie-policy/)中提及的 Cookie 和其他工具。有关我们隐私实践的更多信息，请参阅我们的[隐私政策](https://www.nvidia.cn/about-nvidia/privacy-policy/)。 必备 Cookie 始终有效 这些 Cookie 支持核心功能，如安全性、网络管理和可访问性。它们是网站运行所必需的，不能够关闭。 Cookie 详细信息 性能 Cookie - [x] 性能 Cookie 这些 Cookie 用于量化分析我们网站的访问者，例如您访问的次数、页面停留时间、鼠标移动、滚动、点击和按键活动；其他浏览、搜索或产品研究行为；以及您访问我们网站的原因。它们可以存储唯一 ID，以便我们的系统在您再次访问时能记得您。这些 Cookie 收集的信息用于评估网站性能和寻找改进方法。 Cookie 详细信息 个性化 Cookie - [x] 个性化 Cookie 这些 Cookie 收集有关您如何与我们的网站互动的数据，以帮助我们改善您的网络体验，例如您访问了哪些页面。它们可以存储唯一 ID，以便我们的系统在您再次访问时能记得您。这些 Cookie 可能由我们自身或其服务已添加到我们网页的第三方提供商设置。有了这些 Cookie，我们能够提供增强的网站功能和个性化，并且可让我们发送给您的营销信息更符合您的兴趣。如果您不允许使用这些 Cookie，则部分或全部服务可能无法正常运行。 Cookie 详细信息 广告 Cookie - [x] 广告 Cookie 这些 Cookie 记录您访问我们网站的情况、您访问的页面以及您所浏览的链接，以影响您在其他网站上看到的广告。这些 Cookie 及其收集的信息可能由其他公司（包括我们的广告合作伙伴）管理，并且用于构建您的兴趣画像以及在其他网站上向您展示相关广告。我们与广告合作伙伴将利用这些信息，使我们的网站以及网站上显示的广告更贴合您的兴趣。 Cookie 详细信息 Cookie 列表 Clear * - [x] checkbox label label Apply Cancel Consent Leg.Interest - [x] checkbox label label - [x] checkbox label label - [x] checkbox label label 全部拒绝 保存并接受 [![Image 8: Powered by Onetrust](https://www.nvidia.cn/assets/cn/onetrust/consent/1984719c-04aa-48de-8c22-5174bdf5c0ed/01917434-8292-70f9-8fae-1602521d0b2a/logos/static/powered_by_logo.svg)](https://www.onetrust.com/products/cookie-consent/)