优步(UBER.US)与荷兰云基础设施公司Nebius(NBIS.US)计划共同投入最多3.75亿美元，用于发展Nebius 旗下自动驾驶汽车子公司Avride。Avride发言人表示，这笔新资金(包括战略投资与商业承诺)将有助于 加速Avride自动驾驶车队的扩张，规模预计可达500辆。如果Avride达到双方约定的特定里程碑，两家公 司将进一步追加投资，使潜在总投资额达到3.75亿美元。该公司还表示，这笔投资还将支持Avride的产 品开发及其向新市场的拓展。 优步首席执行官Dara Khosrowshahi此前表示，这家网约车巨头将继续投资自动驾驶技术生态系统，押注 于"人类驾驶与无人驾驶并行"的未来。迄今为止，优步已向多家自动驾驶及无人机公司投资或承诺资 金，包括文远知行(WRD.US)、小马智行(PONY.US)、Lucid Group(LCID.US)、Nuro和Flytrex等，以及运 营或拥有无人驾驶汽车的车队公司。其中一些交易金额高达数亿美元，例如与Lucid、Nuro之间的合 作。 自2020年出售自有自动驾驶部门后，优步已与十多家自动驾驶公司建立合作关系，以与 Alphabet(GOOG ...

Core Insights - The event highlighted the importance of artificial intelligence (AI) and robotics as key drivers of social progress in the context of emerging industries and future technologies [1][3] - The lecture by Academician Zhang Hong focused on the application of AI and robotics in navigation and operational tasks, emphasizing the integration of these technologies into daily life [3][5] Group 1: Event Overview - The 54th session of the Academician Lecture was held in Shenzhen, attracting over 200 participants from various fields including AI and robotics [1] - The event was organized by the Nanshan District Committee and supported by multiple local institutions [1] Group 2: Key Technologies and Applications - Zhang Hong emphasized the goal of enabling robots to operate in open environments using AI technologies, showcasing research on delivery robots and four-legged robotic dogs [3][5] - AI foundational models, such as large language and vision models, provide robots with a "prior knowledge base" to understand their environment and execute tasks based on user commands [5] Group 3: Industry Collaboration and Future Prospects - The event included a "Academician Salon" where over 30 companies, investment institutions, and universities discussed collaboration opportunities in areas like social navigation and AI model empowerment [7] - The ongoing efforts aim to bridge the gap between research and practical application, fostering deep integration of industry, academia, and research for innovation and talent development [7]

9月13日，在北京首钢园举行的2025年中国国际服务贸易交易会上，交通与出行类内容受到关注。图为 观众参观外卖配送机器人。中新社记者 贾天勇 摄 9月13日，在北京首钢园举行的2025年中国国际服务贸易交易会上，交通与出行类内容受到关注。图为 小朋友参观68000吨系列多用途纸浆船首制船"GREEN KEMI"轮模型。中新社记者 贾天勇 摄 9月13日，在北京首钢园举行的2025年中国国际服务贸易交易会上，交通与出行类内容受到关注。图为 小朋友参观一款正向研发的氢能干线物流重型卡车模型。中新社记者 贾天勇 摄 9月13日，在北京首钢园举行的2025年中国国际服务贸易交易会上，交通与出行类内容受到关注。图为 小朋友体验模拟驾驶。中新社记者 贾天勇 摄 9月13日，在北京首钢园举行的2025年中国国际服务贸易交易会上，交通与出行类内容受到关注。图为 观众参观油电混动货运无人机。中新社记者 贾天勇 摄 9月13日，在北京首钢园举行的2025年中国国际服务贸易交易会上，交通与出行类内容受到关注。图为 北京地铁的引导机器人与观众打招呼。中新社记者 贾天勇 摄 ...

五大场景展示具身智能长程、灵巧能力突破 由灵初自研的分层端到端VLA模型与强化学习算法体系提供驱动支撑，本次灵初智能展出的场景Demo 均源于真实客户需求与落地任务验证，覆盖长程复杂任务、开放场景操作、通用泛化性抓取等具身智能 落地过程中的关键挑战： 图注：灵初智能亮相WAIC 2025 图注：灵初智能亮相WAIC 2025 1. 麻将机器人：开放环境下的长程复杂任务 灵初智能麻将机器人与现场观众同桌博弈的场景吸引了众多参观者围观。机器人可与现场观众共同完成 30分钟以上的连续麻将对局，以毫米级精度完成摸牌、放置、碰杠等复杂操作，在与人类玩家的实时博 弈中展现策略思考，动态构建决策链，自主完成碰杠等博弈决策。观众不仅可以现场观看，更能亲自上 桌与机器人对弈，体验机器人如何融合视觉、语言与动作信息，形成游戏博弈和具身操作的融合，展示 真正意义上的"会思考的手"。 图注：机器人与现场观众共同完成30分钟以上的连续麻将对局 图注：机器人自行抓牌和打牌、完成碰杠等博弈决策 2. 自主打包任务：胜任商品级泛化、工具使用、柔性物体操作 从博弈场景转向实用应用，展会现场的零售打包演示区展现了灵初智能在商业场景中的技术实力。 ...

Core Viewpoint - Lingchu Intelligent has made significant advancements in the field of embodied intelligence, showcasing its comprehensive technology chain from data, model algorithms, hardware to application scenarios at WAIC 2025 [2][13]. Group 1: Product Demonstrations - The Mahjong Robot demonstrated the ability to engage in a 30-minute continuous game with human players, showcasing its strategic thinking and decision-making capabilities in a complex environment [2][3]. - The Autonomous Packing Task showcased the robot's ability to understand natural language commands and autonomously execute tasks such as analyzing item arrangement and completing packaging without manual intervention [4]. - The Delivery Robot addressed the "last mile" delivery challenge by accurately identifying and handling various delivery items, demonstrating its flexibility and automation capabilities [6][7]. Group 2: Core Technologies - The 21-degree-of-freedom dexterous hand was highlighted for its exceptional operational dexterity and efficiency, capable of high-precision control and tactile feedback, meeting industrial demands for fine operations [9]. - The Exoskeleton Remote Operation solution was introduced, allowing users to control robotic hands with high precision, contributing to a high-quality data collection platform for reinforcement learning [11]. Group 3: Industry Positioning - Lingchu Intelligent has established a leading position in the L3 long-range dexterous operation sector, marking a significant transition from simple action execution to cognitive decision-making and complex task handling [13]. - The company has built a robust technology ecosystem consisting of four pillars: a data pyramid combining simulation and real data, the VLA model based on the CoAT framework, a cost-effective hardware system, and comprehensive validation across various scenarios [13].