12月5日，2025（第三届）具身智能人形机器人场景应用生态年会在安徽合肥启幕，会议由合肥市包河 区人民政府、移动机器人产业联盟等主办，汇聚60余家供应链企业、40余家产业链企业及千余家行业代 表，聚焦具身智能融合创新与场景落地，共探技术落地与生态共建。 作为网易旗下人工智能研究机构，网易伏羲自成立以来始终致力于利用前沿人工智能技术释放劳动者的 生产力，探索人机协作的全新时代。面对机遇与挑战，网易伏羲以"虚实融合"为核心重塑具身智能生产 力：以有灵智能体激活人机协作效能，以游戏AI沉淀技术反哺实体。 大会期间，网易伏羲凭借突出的技术创新与前沿探索荣膺"偃师·场景应用灵智奖"。"偃师·场景应用灵智 奖"聚焦在技术研发与场景落地双维度表现卓越的企业，为具身智能与机器人行业树立了可借鉴的实践 范本。该奖项旨在激励广大从业者/企业深耕技术创新、勇攀行业高峰，携手构建开放协同、互利共赢 的产业生态，助力智能机器人更深度地服务于社会发展与民生福祉。 此外，网易伏羲大模型与具身智能产品负责人吴雨农受邀分享了《虚实融合：构建具身群智"新"大脑》 主题演讲，系统介绍了网易伏羲在具身智能领域的实践路径与技术突破。 从"Demo" ...

Core Viewpoint - The launch of the "Lingjue" model by NetEase Lingdong represents a significant advancement in autonomous operations for open-pit mining, aligning with national goals for intelligent and unmanned coal mining by 2025 [1][12]. Group 1: Technological Innovations - "Lingjue" is the world's first end-to-end engineering machinery embodiment model, revolutionizing traditional development methods [6]. - The model utilizes real mining operation data for training, overcoming challenges faced by simulation data [6]. - The technology is fully domestically developed, ensuring control over core algorithms and hardware, which enhances security and supply chain stability [6]. Group 2: Operational Efficiency - In harsh conditions at the Hohhot North open-pit coal mine, "Lingjue" achieves an 80% efficiency rate compared to manual loading, with nearly 70% of operation time requiring no human intervention [8]. - The technology significantly improves loading precision and continuity, providing new pathways for safety and efficiency in the industry [8]. Group 3: Industry Collaboration - NetEase Lingdong announced the open-sourcing of the "Lingjue" dataset and initiated the "2027 Industry Collaboration Plan" to promote joint research and standard setting across the industry [9]. - The plan aims to achieve unmanned operations in over 30 mines by 2027 through collaborative efforts with major manufacturers and mining companies [9]. Group 4: Expanding Intelligent Boundaries - The "Mechanical Zhixin" framework, which supports "Lingjue," integrates video learning, expert demonstration, and reinforcement learning, enabling dynamic evolution of machines [10]. - This framework has successfully transitioned to over 10 different scenarios, with future applications planned in agriculture and smart manufacturing [10]. Group 5: Global Leadership in Smart Solutions - NetEase Lingdong has collaborated with over 30 leading enterprises to address labor shortages and safety risks in the industry [12]. - The release of the "Lingjue" model marks a shift from remote control to fully autonomous operations in engineering machinery, positioning China as a leader in global industrial intelligence [12].

Core Insights - The article highlights the transition of industrial AI from cognitive capabilities to autonomous execution, exemplified by the operation of unmanned excavators at the 2025 World Artificial Intelligence Conference (WAIC) [1][2][3] Group 1: Industrial AI Development - Industrial AI is undergoing a significant transformation, showcasing high autonomy and efficiency in complex tasks, such as loading operations in harsh environments [1][2] - The emergence of "Lingju," a model developed by NetEase for unmanned excavators, demonstrates the shift towards end-to-end integrated models that enhance performance and adaptability [2][4] - The collaboration of multiple intelligent agents in industrial settings is redefining production efficiency and flexibility, reducing operation times from hours to minutes [3][5] Group 2: Technological Innovation and Ecosystem - The rise of industrial AI in China is supported by self-innovated foundational technologies, emphasizing the importance of security and stability in the supply chain [3][4] - NetEase's "Lingju" is built on a domestically developed framework, ensuring technological safety and supply chain reliability [4] - Initiatives like the "2027 Industry Collaboration Plan" aim to achieve unmanned operations in over 30 mines by 2027, showcasing a commitment to technological diffusion and ecosystem collaboration [4][5] Group 3: Future of Industrial AI - The concept of a "universal team" of industrial AI agents is emerging, with each agent specializing in different tasks while collaborating to solve complex challenges [5] - The potential for industrial AI to expand into various sectors, including agriculture and smart manufacturing, indicates a redefinition of traditional industry boundaries [5] - The deep integration of AI with the real economy is seen as a key pathway for developing new productive forces [5]

Core Insights - The World Artificial Intelligence Conference (WAIC) witnessed the launch of "Lingjue," the world's first embodied intelligence model specifically designed for open-pit mining excavators, aligning with China's strategic goal for intelligent continuous operations and unmanned transportation by 2025 [1][2] - "Lingjue" represents a significant innovation in engineering machinery, utilizing a unique end-to-end model that leverages multi-modal data-driven autonomous learning technology, marking a departure from traditional modular development [1] - The model has achieved an 80% efficiency rate compared to human operators and requires minimal human intervention, adapting successfully to harsh environments [1] Company Initiatives - NetEase Lingdong announced the open-sourcing of the "Lingjue" dataset and initiated the "2027 Industry Collaboration Plan," aiming to achieve unmanned operations in over 30 mines by 2027 through collaboration with major manufacturers and mining companies [2] - The foundational training framework for "Lingjue," named "Mechanical Wisdom," integrates video learning, expert demonstration, and reinforcement learning, enabling dynamic evolution of machines [2] - The release of "Lingjue" and "Mechanical Wisdom" signifies a shift in engineering machinery from remote control to fully autonomous operations, addressing challenges related to labor shortages and safety risks [2]