Rohan Paul@rohanpaul_ai · 5月24日40

🇨🇳 China's Hangzhou Airport is now using its first track-guided bird-dispersion robot. Has directional sound devices, insect-killing lamps & cameras. Gives runways 24/7 protection with smart patrols, HD cameras, and a greener way to keep birds away.

译🇨🇳 中国杭州机场现已启用其首台轨道式驱鸟机器人。 配备定向声波装置、杀虫灯和摄像头。 通过智能巡逻、高清摄像头和更环保的驱鸟方式，为跑道提供全天候保护。

Rohan Paul@rohanpaul_ai · 5月24日22

Robotic Companies in the United States

译美国机器人公司

Rohan Paul@rohanpaul_ai · 5月24日36

Humanoin in Shenzhen, China. Real-time stability management is among the toughest problems in developing reliable legged robots outdoors.

译中国深圳的Humanoin。 实时稳定性管理是开发可靠的户外双足机器人面临的最棘手问题之一。

Rohan Paul@rohanpaul_ai · 5月24日47

Fei-Fei Li ( @drfeifei ) beautifully explains Robotics. She defines robotics not by form, like humanoids or cars, but by function: they are any "embodied machines" that must perceive, understand, and act within a physical, 3D space. This core requirement is "spatial intelligence," the unifying principle of all robotics, allowing them to perform tasks and even collaborate with humans. Throughout all of human history, we have been confined to a single, shared reality: the "physical Earth 3D world." This singularity has been our only playground. However, new technologies that combine 3D generation and reconstruction are shattering this limitation. We can now create "infinite universes"—a multiverse of digital worlds for countless purposes, from training robots to enabling creativity, travel, and storytelling. This leap from one physical world to an infinite multiverse unlocks boundless possibilities for human imagination and interaction. Video from @a16z

译李飞飞重新定义机器人学，强调其核心是“空间智能”——即机器在三维物理空间中感知、理解与行动的能力。这一能力使机器人能执行任务并实现人机协作。3D生成与重建技术正打破人类仅能体验单一物理世界的局限，创造出用于训练、创造、旅行与社交的无限数字多元宇宙。未来，人们将以“多元宇宙”的方式生活，极大拓展人类想象与交互的边界。

Rohan Paul@rohanpaul_ai · 5月23日25

Dyson has deployed robotic arms that selectively harvest strawberries based on ripeness detection in their innovative vertical farming system located in the UK.

译戴森在其位于英国的创新垂直农场系统中，部署了能够根据成熟度检测选择性采摘草莓的机械臂。

Rohan Paul@rohanpaul_ai · 5月22日46

Unitree Robotics' G1 humanoid robot playing table tennis against a human during a public exhibition demo at the Global Unicorn Innovation Exhibition in Hangzhou, China.

译宇树机器人公司的G1人形机器人在中国杭州全球独角兽创新展览的公开演示中与人类进行乒乓球对打。

Rohan Paul@rohanpaul_ai · 5月22日46

This RAI Institute robot managing 3-balls juggling through dynamic hand adjustments. It processes visual and contact information to maintain the pattern without external aids.

译这个RAI研究所的机器人通过动态手部调整管理三球抛接。它处理视觉和接触信息以维持模式，无需外部辅助。

Rohan Paul@rohanpaul_ai · 5月22日32

Edge AI runs on each insect backpack, enabling low-latency coordination, secure data exchange, formation control as a group and task execution. Swarm Biotactics scales by breeding insects and has raised ~€13M.

译边缘AI运行在每只昆虫的背包上，实现低延迟协调、安全数据交换、群体编队控制和任务执行。 Swarm Biotactics通过培育昆虫实现规模化，已融资约1300万欧元。

Baidu Inc.@Baidu_Inc · 5月22日21

Couldn't catch Baidu Create 2026 this year? Come along as we explore our newest AI products at our exhibition, including a DuMate-powered robot that holds its own at mahjong.

译今年没赶上百度Create 2026？来和我们一起逛展，看看我们最新的AI产品，包括一个能独立打麻将的DuMate机器人。

Berryxia.AI@berryxia · 5月22日57

Optimus V2.5 走路的样子已经明显变了。 视频里它迈步时有了清晰的节奏和自信，动作连贯自然，不再像之前那样带着明显的机械感和谨慎。 这个进步不是小事。 行走一直是人形机器人最难解决的动态平衡问题之一。 现在它能走得像一个真正知道自己要去哪里的人，说明整个感知、控制和执行系统的协同能力又上了一个台阶。 当 Optimus 连走路都已经开始带上人的姿态时。 我们真正该关注的，已经从它能不能走稳，变成了它什么时候能真正进入工厂、仓库和家里开始干活。

译Tesla Optimus V2.5的行走动态展现出显著提升，动作更连贯、自然，充满自信。这一进步反映了其感知、控制与执行系统的协同能力达到了新高度，解决了人形机器人动态平衡的核心难题。讨论焦点已从其能否走稳，转向何时能真正进入工厂、仓库等实际场景工作。

AYi@AYi_AInotes · 5月22日14

holy shit，确定这不是AI吗？ AI标志在哪？ 如果是真的，Tony老师会不会失业？？

译我的天，这确定不是AI吗？ AI标志在哪里？ 如果是真的，Tony老师会不会失业？？

Berryxia.AI@berryxia · 5月21日46

牛逼！ 中国🇨🇳可以吃上正餐了！ FSD在中国要落地了～

译太棒了！ 中国🇨🇳可以吃上正餐了！ FSD在中国要落地了～

小互@xiaohu · 5月21日78

FSD来了… 官宣进入大陆…

Chubby♨️@kimmonismus · 5月21日28

„We are now talking about physical AGI, and define it as all of what humans can do.“ Really interesting developments on robotics as well. Will cover everything later

译我们现在讨论的是物理AGI，并将其定义为人类能做的一切。 机器人领域也有非常有趣的进展，稍后将全面介绍。

SemiAnalysis@SemiAnalysis_ · 5月21日58

The full chat with Mishek Musa on how ADI is shrinking inference down to the edge and setting up physical leaderboards for the robotics community. Chapters: 0:00 — Introduction & ADI's Emerging Tech Hub 0:56 — Inside the Multimodal Tactile Sensor 1:52 — Automating Data Center Maintenance 2:28 — Open-Source Robotics Benchmarks 3:24 — High-Fidelity Simulation Assets 4:00 — The System-Level Product Strategy 4:37 — Data Collection & Minimizing the Sim-to-Real Gap 5:53 — Co-Innovation Hub Collaborations 6:30 — Distilling Large Models for Edge Inference 7:47 — Custom Co-Designed Silicon vs. Generic GPUs 8:59 — Wrap-Up & Concluding Thoughts

译ADI正在展示其将大型AI模型能力从云端下沉到边缘设备的技术路径，核心是通过模型蒸馏、定制化协同设计芯片等手段实现高效推理。同时，ADI正为机器人社区构建开源的基准测试与物理排行榜，并致力于开发多模态触觉传感器、高保真仿真资产等，以最小化仿真与现实之间的差距。这体现了其从系统层面推动硬件协同创新与数据采集的生态化产品战略。

小互@xiaohu · 5月21日39

由Gemma 4 驱动的 Open Duck 机器人 有视觉能力，还能对话

译由Gemma 4驱动的Open Duck机器人 具备视觉能力，还能对话

Rohan Paul@rohanpaul_ai · 5月21日50

Chinese startup Rochu Robotics developed a humanoid hand that mimics real anatomy using hydraulics and 24 biomimetic tendons for flexible, lifelike motion. Features a one-to-one skeletal design to move more like a real human hand.

译中国初创公司 Rochu Robotics 开发了一款仿生机械手，它通过液压系统和24条仿生肌腱模拟真实手部解剖结构，实现灵活、逼真的动作。 其采用一对一骨骼设计，使其运动更接近真实人手。

AYi@AYi_AInotes · 5月21日16

Damn，机器人跳着跳着《Billie Jean》就瘫倒了😭

译该死，机器人跳着跳着《Billie Jean》就瘫倒了😭

AK@_akhaliq · 5月21日64

ESI-Bench Towards Embodied Spatial Intelligence that Closes the Perception-Action Loop

译ESI-Bench 迈向闭环感知-行动的具身空间智能

DogeDesigner@cb_doge · 5月20日58

Grok Summary of Elon Musk's Forbes interview from today. OpenAI Lawsuit & Verdict Musk called the verdict a “dangerous precedent”. He argued that allowing a nonprofit to convert into a for-profit (especially after removing key protective clauses) undermines charitable giving in America. He described the jury’s decision as dubious because it overlooked the gradual nature of the conversion and plans to appeal to establish stronger protections against what he sees as “looting” charities. AI Predictions & Timeline Musk painted a picture of extremely rapid progress: •AI breakthroughs are happening constantly (“When I go to sleep, there’s an AI breakthrough; when I go to lunch, there’s a breakthrough”). •In ~5 years, digital intelligence could exceed the sum of all human intelligence. •The global economy may roughly double in size within 5–7 years. •Humanoid robots: At least 100 million in 5 years, potentially up to a billion. •AI is already “vastly smarter than humans” in some domains; he hopes it will be “nice to us.” He emphasized that AI compute (especially for training and inference) will increasingly move to space because of abundant solar power and the ability to scale without Earth-based grid or land constraints. SpaceX & Multi-Planetary Future Musk reiterated SpaceX’s core mission: making humanity multi-planetary as a backup for civilization. He highlighted progress toward fully reusable rockets (targeting major capability by year-end) that could enable massive cargo shipments (millions of tons) to the Moon and Mars to build self-sustaining cities. He also touched on the value of the existing Starlink satellite constellation for future space-based infrastructure, including potential orbital data centers. Neuralink & “Jesus-Level” Tech Musk described Neuralink’s brain-machine interfaces as capable of delivering near-miraculous outcomes — restoring eyesight, mobility, and speech for people with disabilities. He framed these as high-priority “Jesus level” innovations that directly extend and improve human capability. Other Big Ideas & Untapped Opportunities Musk pointed to several areas ripe for disruption: •Tunnels — 3D transportation networks to eliminate surface traffic (he encouraged others to start tunnel companies). •Synthetic/digital medicine — Custom RNA and related technologies that could effectively “cure anything.” •Electric aircraft and other sustainable transport. •Space-based AI infrastructure — Leveraging solar power for massive compute clusters. Legacy & Mindset When asked what he wants to be remembered for in 250 years, Musk replied simply: “He played a useful role in the advancement of civilization.” His focus remains on the technologies needed to extend life beyond Earth and accelerate human progress. He named Nikola Tesla as a top historical inspiration and Jensen Huang among current ones. Overall tone: Classic Musk — zero victimhood about the OpenAI loss, maximum forward-looking vision, rapid topic shifts, and a sense of urgency about AI, space, and extending civilization. The interview blends candid legal criticism with sweeping predictions about a future of abundant energy, intelligent machines, and humanity becoming multi-planetary.

译在《福布斯》访谈中，埃隆·马斯克就多个领域阐述了激进观点。他批评针对OpenAI的诉讼败诉开创了“危险先例”，并计划上诉。其核心预测包括：AI发展呈指数级，5年内数字智能或超全人类智能总和；全球经济规模有望数年内翻倍；人形机器人将达数亿台。SpaceX致力于开发全复用火箭，以实现大规模太空运输并建立地外城市。他将Neuralink脑机接口技术视为“耶稣级”创新，能恢复残障人士机能。此外，他还提及了隧道交通、合成医学等机遇，整体展现出以技术加速人类文明进程的强烈紧迫感与乐观构想。

Rohan Paul@rohanpaul_ai · 5月20日44

🇨🇳In Shenzhen China, food glides into your table mid-air, guided by AI. Delivery pods use magnetic levitation, AI routing, and linear motors for smooth, wheel-free motion. Each pod maps the space, avoids collisions, and optimizes routes in real time

译🇨🇳在中国深圳，食物通过空中滑行送到你的餐桌，由AI引导。 配送舱使用磁悬浮、AI路径规划和直线电机，实现平稳、无轮的运动。每个配送舱实时绘制空间地图、避开碰撞并优化路线。

DogeDesigner@cb_doge · 5月20日41

ELON MUSK: Brain chips could create “Jesus-Level” Miracles. - Brain-machine interfaces could give people cybernetic superpowers. - Neuralink could help people with brain or spine injuries speak again, see again, and even walk again. - Direct brain interfaces may restore eyesight for people who lost both optic nerves, or even those born blind. - These breakthroughs feel like “Jesus-level miracles” because they could change human lives in a profound way.

译埃隆·马斯克：脑芯片或能创造“耶稣级”奇迹。 - 脑机接口或能赋予人类赛博格超能力。 - Neuralink或可帮助脑部或脊髓损伤患者重新说话、视物，甚至行走。 - 直接脑接口或可为双眼视神经受损者，甚至先天失明者恢复视力。 - 这些突破如同“耶稣级奇迹”，因其可能深刻改变人类生活。

DogeDesigner@cb_doge · 5月19日42

ELON MUSK: "In 5 years, digital intelligence will exceed the sum of all human intelligence." Within five years, there may be at least 100 million humanoid robots, possibly even 1 billion. The economy could double in size within 5 to 7 years because AI and robotics may increase output dramatically. The pace of change will be so fast that the world could look very different in just a few years.

译埃隆·马斯克：“5年内，数字智能将超越人类智能总和。” 五年内，人形机器人数量可能至少达到1亿，甚至可能达到10亿。 由于AI和机器人技术可能大幅提升产出，经济规模或在5到7年内翻倍。 变化速度将如此之快，短短几年内世界可能面貌全非。

DogeDesigner@cb_doge · 5月19日29

ELON MUSK: "Building the technology is necessary to extend life beyond earth. There’s the Starlink internet, which is rebuilding the entire internet in space. I guess that’s kind of pretty cool. There’s Optimus robot that we’re developing at Tesla, the self-driving cars."

译埃隆·马斯克："构建技术对于将生命扩展到地球之外是必要的。有星链互联网，它正在太空中重建整个互联网。我想这挺酷的。还有我们在特斯拉开发的Optimus机器人，以及自动驾驶汽车。"

Rohan Paul@rohanpaul_ai · 5月19日58

AI leaving screens and becoming useful in places where objects, people, shelves, and sensors interact in real time. Radar is building the perception layer for retail that can turn messy stores into machine-readable environments where AI can identify, locate, and reason about products in real time. It is basically the “operating system for physical stores”. A lot of times, physical stores do not know whether a shirt is in the backroom, under another pile, in the wrong aisle, or if it’s already stolen. Radar fixes that by giving stores a live map of inventory, down to about 10cm, so a worker can find the exact item instead of guessing. The smart part is the hybrid design. Cameras can see shelves and movement, but crowded racks confuse them. RFID tags can identify items, but they need spatial awareness to know what is actually happening around them. Radar combines both, so the store gets the identity of each product plus the visual context around it.

译RADAR正通过融合摄像头与RFID的混合感知技术，打造“实体店的操作系统”。该系统能将实体零售环境转化为机器可读空间，提供精度达10厘米的实时库存地图，解决长期困扰行业的库存可视化难题。公司近期完成1.7亿美元B轮融资，估值突破10亿美元，其Physical AI技术已在超1400家门店部署，实现99%的单品级实时库存准确率，致力于弥补实体零售因库存不透明导致的每年约万亿美元损失。

Rohan Paul@rohanpaul_ai · 5月19日71

Humanoid value will not come from looking human, but from having enough body surface, strength, balance, and feedback to turn messy objects into manageable ones.

译人形机器人的核心价值不在于外形相似，而在于具备足够的物理能力（如力量、平衡和全身协调）来处理复杂任务。实现这一目标的关键是“全身控制”，即机器人能调动全身与环境互动并适应负载变化。波士顿动力的Atlas机器人通过本体感知成功处理超过100磅的动态负载，展示了这种能力。为实现高性能操作，团队已放弃传统MPC控制范式，全面转向强化学习（RL）。这种全身控制能力是物理智能的基础，也是人形机器人价值主张的核心。

歸藏(guizang.ai)@op7418 · 5月19日48

波士顿动力机器人的新演示，现在可以搬动很重的东西

AYi@AYi_AInotes · 5月18日26

Damn，波士顿机器人这么牛逼了， 家政、搬运工不得失业一大批😭

Rohan Paul@rohanpaul_ai · 5月18日58

Boston Dynamics showed Atlas lifting and carrying a 100+ lb mini-fridge, using reinforcement learning to handle weight, grip, position, and balance through body proprioception. shows how humanoids may handle hard labor: not by seeing objects better, but by adapting through contact, body feedback, domain-randomized training, and hardware built for strength and repairability.

译Boston Dynamics展示了Atlas机器人使用强化学习搬运超100磅小冰箱，通过全身感知协调处理重量、抓握与平衡。这体现了人形机器人处理重体力任务的核心逻辑：不依赖视觉识别，而是通过接触适应、本体感知反馈、针对特定领域的随机化训练，以及专为力量与可维护性设计的硬件来完成复杂协作。引用的背景信息进一步说明，Atlas已能精准可靠地协调全身关节，管理重型物体的复杂接触点。

AYi@AYi_AInotes · 5月18日50

holy shit，连他么快递分拣员都要失业了吗？人类早晚玩完🤣

译我的天，连快递分拣员都要失业了吗？人类迟早完蛋🤣

DogeDesigner@cb_doge · 5月18日52

Grok Summary of Elon Musk’s interview at the Samson International Smart Mobility Summit today. 1. Strong praise for Israel’s innovation edge
Elon called out Israel’s outsized impact, saying it “punches way above its weight, probably #1 in the world” in innovation per capita. He expressed clear admiration for the country’s tech ecosystem, especially in AI and mobility. 2. Tesla FSD & unsupervised robotaxis — near-term reality •Unsupervised robotaxis are coming soon. •Tesla is already running driverless tests in select Texas areas. •FSD availability: Elon stated it will roll out in both the US and Israel by the end of 2026. •As the AI improves, cars will feel increasingly “alive.” •Prediction: In about 10 years, most driving will be AI-handled. 3. Humanoid robots → abundance economy
Elon painted a future where humanoid robots (Optimus and others) vastly outnumber humans. This shift, he said, will create massive abundance and could enable ideas like “universal high income.” He framed smart mobility as part of this broader robotics/AI revolution. 4. Starship, multi-planetary life & Neuralink •Starship’s rapid reusability is the key unlock for becoming multi-planetary and building cities on Mars. •Neuralink was briefly highlighted for restoring function to people with paralysis or vision loss — described in one summary as potentially “Jesus-level” impact. •All of these threads (autonomy, robots, space, brain interfaces) connect to one overarching goal: “maximize the probability that civilization has a great future.” 5. Balanced optimism with risk awareness
Classic Elon: big-picture excitement tempered by realism. He noted risks (including rogue robots/AI) but emphasized proactive development and deployment as the path forward.

译在智能出行峰会上，Elon Musk展望了由AI和机器人驱动的未来。他透露，特斯拉完全自动驾驶系统（FSD）及无人驾驶出租车业务预计将于2026年底前在美国和以色列推出。他构想人形机器人将远超人类数量，创造巨大物质丰富，甚至可能实现“普遍高收入”。Musk强调，星舰的快速可复用性是实现火星殖民的关键，而Neuralink等脑机接口技术旨在恢复残障人士的功能。整体上，他平衡了乐观与风险意识，认为自主技术、机器人、太空探索与脑机接口共同致力于提升文明未来的概率。

DogeDesigner@cb_doge · 5月18日69

ELON MUSK: "Tesla full-self driving software, which is really just AI and cameras - we don't use radars or lidar or anything like that. It's really trying to drive the car in the same way that human drives the car, which is human drives the car primarily with vision and with a biological neural net, that we take the same approach with our vehicles, which is a digital neural net, and cameras, and we expect this approach to ultimately be at least an order of magnitude safer than humans driving. It is quite magical, because the car feels like it is sentient. It feels actually feels like it's alive. We already have some vehicles operating with no people inside and no safety monitors in three cities in Texas, and probably will be widespread in the US."

译马斯克阐述Tesla全自动驾驶（FSD）软件完全基于AI与摄像头，不使用雷达或激光雷达，通过数字神经网络模仿人类以视觉为主驾驶车辆的方式。他预期该技术最终将至少比人类驾驶安全一个数量级，并形容车辆表现得仿佛具有知觉。目前FSD已在德州三个城市实现无安全员运营，预计将在美国广泛普及。

DogeDesigner@cb_doge · 5月18日44

ELON MUSK: "My prediction is that there'll be far more robots, like intelligent robots, in the world than there will be people, and I think this is most likely to be a good thing, we always want to be a little paranoid, or certainly not complacent about the safety of robots, but I think it will usher in an age of not universal basic income, but universal high income."

译埃隆·马斯克：“我的预测是，世界上智能机器人的数量将远超人类，我认为这很可能是一件好事。我们总是希望对机器人安全保持一点警惕，或者至少不能掉以轻心，但我认为这将开启一个不是全民基本收入，而是全民高收入的时代。”

meng shao@shao__meng · 5月18日13

Figure AI 这个 PR 视频，槽点太多，感觉甚至不如去跑马拉松 😂

小互@xiaohu · 5月18日72

Figure 直播机器人 VS 人类 挑战 快递分拣工作 目前人类稍稍领先…😌

译Figure 直播机器人 VS 人类 挑战 快递分拣任务 目前人类稍稍领先…😌

Berryxia.AI@berryxia · 5月18日54

大佬永远比普通人站的更高，看的更远！ Yann LeCun最近又放出重磅预测。 这位Meta AI首席科学家、图灵奖得主、现代计算机视觉之父，直接说：12到18个月内，我们就会有通用方法来训练分层世界模型。 这些模型会直接从视频和真实世界数据里学习。 学完就能帮机器人规划动作、帮医疗系统做决策、帮更多领域解决物理世界里的实际问题。 最后一步，是把它扩展成一个通用的世界模型。 大家还在拼命卷LLM的参数和上下文长度，LeCun却把目光放在了真正能理解物理因果、能规划真实行动的世界模型上。 这可能是从“会聊天”走向“会做事”的关键一步。

译Meta AI首席科学家Yann LeCun预测，未来12到18个月内将出现训练分层世界模型的通用方法。这些模型将从视频和真实世界数据中学习，具备理解物理因果和规划行动的能力，可应用于机器人、医疗等多个领域解决实际问题。最终目标是将其扩展为通用的世界模型。这标志着AI研究重点可能从当前以LLM为代表的“会聊天”模型，转向能够理解并作用于物理世界的“会做事”模型。

Chubby♨️@kimmonismus · 5月18日39

Thankfully, people will soon no longer have to do this job.

译值得庆幸的是，人们很快就不再需要从事这项工作。

Rohan Paul@rohanpaul_ai · 5月17日33

Love at first crash. 😍 literally. In Philadelphia, a delivery robot crossing the street took a glancing blow from a truck. Heart eyes appeared on its LED panel immediately after. The bot adjusted and continued its route.

译一见钟情。😍 字面意义上的。 在费城，一个正在过马路的送货机器人被卡车擦撞。随后其LED面板上立即出现了爱心眼睛图案。机器人调整状态后继续执行路线。

Berryxia.AI@berryxia · 5月17日71

哎！这玩意越看让人越有点感慨不已！ 人形机器人的真的逐渐在替代某些岗位啊！ Figure 人形机器人已经进入第4天 nonstop autonomous operations了。 F.03 正在24/7连续自主运行，直到失败为止。 直接怼着真实仓库环境里的耐力测试：机器人自己抓取、搬运、分拣、循环工作，持续收集真实故障数据、维护时机、安全恢复机制和人类监督需求。 这才是人形机器人从“能动”走向“能干”的关键一步。 以前大家看的是单次表演，现在看的是它能不能真正扛住连续工作。 直播还在进行中：https://twitter.com/i/broadcasts/1OxwblMvXvoJB

译Figure公司的F.03人形机器人已进入第四天不间断自主运行测试，在真实仓库环境中24/7连续工作直至出现故障。测试核心在于评估机器人执行抓取、搬运、分拣等任务的长期耐力，并收集故障数据、维护需求及安全恢复机制等信息。这标志着人形机器人从展示单次动作的“能动”阶段，进入了考验持续工作能力的“能干”实用化关键阶段。

Berryxia.AI@berryxia · 5月16日63

兄弟们，具身智能这下真的靠点谱了啊！ 具身智能（Embodied AI）下一个真正的大前沿来了。 HuggingPapers刚刚推送了一篇重磅综述：《World Action Models: The Next Frontier in Embodied AI》 这是第一篇系统定义「World Action Models（WAMs）」的论文。 WAMs 的核心是：同时预测未来世界状态 + 生成真实可执行动作的具身基础模型。 它不再是单纯“想想就行”的语言模型，而是真正能理解物理世界、预测变化、并采取行动的智能体。 论文系统梳理了当前所有WAMs的架构设计、数据生态系统和评估协议，还附了一张2024-2026年的完整发展时间线图，一目了然。 Project page：https://openmoss.github.io/Awesome-WAM/ Paper：https://huggingface.co/papers/2605.12090 如果你在做机器人、具身Agent、物理世界AI或者世界模型，这篇综述来得正是时候。

译HuggingPapers发布首篇系统性定义“世界行动模型”的综述论文。WAMs被视为具身智能的下一个前沿，其核心是能同时预测未来世界状态并生成真实可执行动作的具身基础模型，超越了仅能推理的语言模型。论文系统梳理了WAMs的架构设计、数据生态系统和评估协议，并提供了发展时间线图，对从事机器人、具身Agent、物理世界AI及世界模型的研究者具有重要参考价值。