对话 | 唐小引 嘉宾 | 唐睿 责编 | 梦依丹 出品 | CSDN（ID：CSDNnews） 当大模型开始"看懂"空间、理解物理、做出行动，人工智能的形态正在发生一次根本性变化——从"对话系统"，走向"行动智能"。 在这条路径上，一个词被频繁提起：空间智能。 以下文章来源于CSDN ，作者万有引力 CSDN . 成就一亿技术人 在全球机器学校技术大会现场，唐睿在与 CSDN 《万有引力》栏目的深度对话中，不仅给出了他的答案，更剖 析了行业深处的痛点与机遇。以下是访 谈中唐睿表达的一些观点提炼： 欢迎 收听音频播客，如有兴趣观看完整视频，可在文末获取 以下是对话的完整内容： 唐小引：屏幕前的小伙伴们大家好，欢迎收看《万有引力》。今天我们来到全球机器学习技术大会的现场，特别邀请到了群核科技首席科学家唐睿老 师，和大家一起深入分享他的技术人生成长，还有大家当前很关注的对于空间智能的整个思考、研究以及实践。欢迎我的本家唐老师，可以给大家打个 招呼，然后做一下自我介绍。 如果说 LLM 让机器拥有了像人类一样思考的大脑，那么空间智能则试图赋予机器像人类一样观察、理解并在三维世界中行动的身体与感官。 它并非凭空出现， ...

LingBot-Depth模型依托奥比中光Gemini330系列双目3D相机进行RGB-Depth数据采集与效果验证，并基 于深度引擎芯片直出的深度数据进行训练与优化，实现了空间智能算法的创新突破。值得一提的是， LingBot-Depth模型已通过奥比中光深度视觉实验室的专业认证，在精度、稳定性及复杂场景适应性方 面均达到行业领先水平。 目前，双方已达成战略合作伙伴关系，将基于LingBot-Depth模型推出新一代深度相机，依托Gemini 330 系列相机提供的芯片级3D数据，进一步通过技术协同、生态共建，为机器人处理各行各业极端场景、 真正落地提供强大的技术支撑。 作为行业领先的机器人及AI视觉科技公司，奥比中光在LingBot-Depth模型的研发与测试阶段提供了关 键硬件支持，LingBot-Depth模型已通过奥比中光深度视觉实验室认证。未来，双方将持续展开技术协 同，共同推动空间智能从算法创新走向真实场景落地。 蚂蚁灵波科技作为蚂蚁集团旗下的具身智能公司，专注具身智能基座的研发，本次发布的LingBot- Depth攻克了具身智能在真实复杂环境中面临的视觉感知难题。针对玻璃、镜面、金属等透明或 ...

蚂蚁灵波科技作为蚂蚁集团旗下的具身智能公司，专注具身智能基座的研发，本次发布的LingBot- Depth攻克了具身智能在真实复杂环境中面临的视觉感知难题。针对玻璃、镜面、金属等透明或高反光 物体导致的深度信息缺失，该模型基于奥比中光的Gemini330系列双目3D相机提供的芯片级原始数据， 智能补全深度信息，显著提升机器人在复杂光学场景下的感知鲁棒性与作业成功率。 记者从奥比中光(688322)获悉，1月27日，蚂蚁集团旗下蚂蚁灵波科技发布其首个开源的空间感知模型 LingBot-Depth，并宣布与奥比中光达成战略合作意向。奥比中光在LingBot-Depth模型的研发与测试阶 段提供了关键硬件支持，LingBot-Depth模型已通过奥比中光深度视觉实验室认证。据悉，未来双方将 持续展开技术协同，共同推动空间智能从算法创新走向真实场景落地。 蚂蚁灵波科技CEO朱兴表示，LingBot-Depth验证了"高质量芯片级深度数据+面向真实场景的算法建 模"对提升复杂环境可用性的价值。在不改变既有传感器形态的前提下，通过软硬协同与工程化评测闭 环，有望让更多机器人在透明/反光等高难场景中获得更稳定的深度输入。 ...

LingBot-Depth 的优异性来源于海量真实场景数据。灵波科技采集约 1000 万份原始样本，提炼出 200 万 组高价值深度配对数据用于训练，支撑模型在极端环境下的泛化能力。这一核心数据资产（包括 2M 真 实世界深度数据和 1M 仿真数据）将于近期开源，推动社区更快攻克复杂场景空间感知难题。 空间智能迎来重要开源进展。1月27日，蚂蚁集团旗下具身智能公司灵波科技宣布开源高精度空间感知 模型LingBot-Depth。 该模型基于奥比中光 Gemini 330 系列双目 3D 相机提供的芯片级原始数据，专注于提升环境深度感知与 三维空间理解能力，旨在为机器人、自动驾驶汽车等智能终端赋予更精准、更可靠的三维视觉，在"看 清楚"三维世界这一行业关键难题上取得重要突破。这也是蚂蚁灵波科技在2025外滩大会后首次亮相 后，时隔半年在具身智能技术基座方向公布重要成果。 在NYUv2、ETH3D等权威基准评测中，LingBot-Depth展现出代际级优势：相比业界主流的 PromptDA 与PriorDA，其在室内场景的相对误差（REL）降低超过70%，在挑战性的稀疏SfM 任务中RMSE误差降 低约47%。 在 ...

Core Viewpoint - World Labs, founded by Fei-Fei Li, is seeking to raise up to $500 million at a valuation of approximately $5 billion, marking a significant increase from its previous valuation of $1 billion in 2024, indicating a 5x revaluation in just over a year [2][4]. Financing and Valuation - If the financing is successful, World Labs' valuation will jump from $1 billion to $5 billion, reflecting a rapid increase in investor confidence in its "world model" approach [2][4]. - World Labs has previously raised a total of $230 million, with initial funding rounds led by notable investors such as Andreessen Horowitz and Radical Ventures, and later rounds involving firms like NVIDIA and Temasek [5][6]. Product Development - World Labs is developing AI systems capable of navigation and decision-making in three-dimensional environments, focusing on creating "large world models" that understand the structure and evolution of the physical world [8][9]. - The company launched its first 3D world generation model, Marble, which can create explorable 3D environments based on text or image prompts, utilizing advanced techniques like 3D Gaussian Splatting for efficient rendering [10][14]. Strategic Importance - Fei-Fei Li emphasizes that world models are crucial for achieving spatial intelligence and are considered the next core focus for AI in the coming decade, following large language models [16][18]. - The world model is seen as a foundational capability that can influence multiple application areas, providing predictive representations of environments essential for effective decision-making and control [18][22]. Competitive Landscape - Another significant player in the world model space is AMI Labs, founded by Yann LeCun, which is pursuing a different approach focused on implicit world models. This indicates a broader investment interest in various technological paths within the world model domain [20][24]. - The world model landscape can be categorized into three layers, with LeCun's JEPA positioned at the highest abstract level, highlighting the diverse strategies being adopted by different companies in this field [24][27].

Core Viewpoint - The company, Sitongqi, has reported significant industry achievements in the field of spatial intelligence through its affiliate, Kaola Youran, which is focused on technological innovation and product development in the artificial intelligence sector [1] Group 1 - The affiliate, Kaola Youran, is committed to continuous technological iteration and product innovation [1] - The company aims to become a leading enterprise in technological innovation and industry implementation within the artificial intelligence sector [1]

Core Insights - The Gaode Street Ranking has undergone a comprehensive upgrade, introducing the world's first "Flying Street View" and a dynamic lifestyle service ranking covering all seasons, categories, and demographics [1][4][5] - The platform has achieved significant user growth, with over 660 million users and a 270% increase in merchant revenue since its launch [1][9] - The introduction of a trust-based recommendation mechanism enhances user engagement and interaction, allowing users to see friends' ratings and create their own lists [7][8] Group 1: Product Features - The "Flying Street View" feature allows users to experience a realistic view of stores and surroundings before visiting, addressing common navigation challenges [4][11] - A total of 6,553 seasonal rankings and 1,550 category rankings have been created, showcasing the platform's ability to generate dynamic content based on time and location [5] - The platform will soon introduce an "AR Real Scene" feature, providing users with interactive experiences related to restaurants and tourist attractions [4] Group 2: User and Merchant Impact - The Gaode Street Ranking has attracted 860,000 new merchants, with order volumes increasing by over 330% [9] - The platform's focus on real user behavior data has established a more trustworthy service credit system, enhancing consumer confidence [2][8] - Merchants can now create their own "Flying Street View" at no cost, allowing them to showcase their unique offerings without significant marketing expenses [12] Group 3: Industry Implications - The upgrade signifies a shift in the local service industry towards a more authentic evaluation system, promoting healthy competition among businesses [12][13] - The integration of AI and spatial intelligence positions Gaode as a key player in the evolving landscape of local services, moving beyond traditional navigation tools [14] - The competition in the local service sector is transitioning from price subsidies to a focus on reliable data and AI capabilities [14]

站在2026年的起点，展望全球人工智能（AI）发展，技术、产业、能源、治理多重变量交织，将共同 塑造这一关键年份。 相关机构预测，越来越多的顶尖AI企业将聚焦提升大模型推理能力与智能体执行任务能力，推动AI 从"会生成"向"会规划、会行动"进化。大量企业应用将嵌入任务型AI智能体。 与技术突破相伴的则是能源压力，全球数据中心耗电量将持续高企。治理层面，预计各国治理措施将加 速落地。 技术：大模型竞赛带动智能体应用 2026年，人工智能大模型你追我赶的竞争趋势将延续。开放人工智能研究中心（OpenAI）、谷歌、深 度求索等企业将发布规模更大或效率更高的最新版本大模型。 著名人工智能研究者、美国斯坦福大学教授李飞飞日前撰文指出，空间智能是人工智能下一个前沿。大 模型在成功处理文本数据、多模态数据的基础上，正在空间理解力方面取得进步，其目标是具备语义、 物理、几何、动态复杂交互等方面能力的模型。 同时，智能体可能日益普及，人工智能与人们的生活结合得将更为紧密。传统AI系统工作模式是一问 一答，而具备深度目标导向、更多步骤规划能力以及擅长特定任务的智能体将越来越多地应用于各种工 作中。美国高德纳咨询公司预测，2026 ...

Core Insights - The year 2026 is anticipated to be a watershed moment for embodied intelligence, with humanoid robots gaining attention but facing challenges in real-world applications [1] - The bottleneck in the industry is not the robots' ability to perform tasks but their reliability in unpredictable environments [1][2] - The focus is shifting from traditional methods to a vision-based approach for embodied intelligence, aiming to create a more robust and adaptable system [2][5] Industry Challenges - The current industry anxiety stems from the inability of robots to transition from demonstration to practical use in factories and homes [1] - The long-tail reliability in real-world scenarios poses significant challenges, as environmental factors can disrupt robotic operations [1][3] Technological Innovations - Chenjing Technology is pioneering a vision-based approach, moving away from expensive lidar systems to create a more cost-effective and scalable solution [2][5] - The concept of "spatial intelligence" is introduced, emphasizing the need for robots to have a precise understanding of their physical environment [3][4] - The upcoming product from Chenjing Technology aims to enable robots to build internal world models that are computable, predictable, and executable [4] Product Development - The new product lineup includes the Insight autonomous spatial camera, which utilizes advanced neural network computing and a high-performance VSLAM engine for enhanced perception [8] - The TinyNav high-performance navigation algorithm library is designed to provide robust positioning and mapping capabilities even in low-cost embedded systems [8] - The RoboSpatial toolchain allows developers to easily manipulate 3D spatial intelligence capabilities, streamlining the deployment of embodied intelligence applications [9] Strategic Partnerships - Chenjing Technology has partnered with industry leader Yushu Technology to enhance the robustness of its visual perception solutions in complex environments [7]

Core Viewpoint - The collaboration between World Labs, led by Fei-Fei Li, and Guanglun Intelligent, a leading synthetic data company, aims to address the long-standing issue of "scalable evaluation" in the field of embodied intelligence, marking the entry into an evaluation-driven era for this technology [1][2][3]. Group 1: Companies Involved - World Labs is founded by Fei-Fei Li, a prominent figure in AI, known for her work on ImageNet and as a former chief AI scientist at Google Cloud [4][5]. - Guanglun Intelligent is recognized as a hot company in the embodied intelligence infrastructure sector, having established a strong partnership with NVIDIA and contributing to the development of simulation systems [54][55]. Group 2: Technological Innovations - World Labs is set to launch its first product, Marble, by the end of 2025, which can generate high-fidelity 3D worlds from minimal input [8][9]. - Marble aims to provide a visualized world model, allowing users to create and export 3D environments efficiently, thus serving as a productivity tool for visual effects and game developers [15][16]. Group 3: Challenges in Evaluation - The rapid advancement of models in embodied intelligence has outpaced existing benchmarks, creating a need for new evaluation methods [20][22]. - Traditional evaluation methods are inadequate for assessing the capabilities of embodied intelligence, necessitating the use of simulation as a scalable solution [29][30]. Group 4: Strategic Collaboration - The partnership between World Labs and Guanglun Intelligent is crucial for developing a comprehensive evaluation framework that integrates environment generation and physical interaction [37][49]. - Guanglun Intelligent's role is to provide the necessary physical assets and evaluation loops, ensuring that the simulated environments can support real physical interactions [49][50]. Group 5: Future Directions - The collaboration signifies a pivotal moment in the embodied intelligence sector, as it transitions into an evaluation-driven era, with the potential to shape research directions and identify technological bottlenecks [71][72][76]. - The establishment of robust evaluation standards, such as RoboFinals, highlights the industry's shift towards scalable and credible assessment frameworks for advanced robotic models [63][64].