Rohan Paul@rohanpaul_ai · 4月19日

In Beijing's 2026 humanoid robot half-marathon, HONOR's Lightning completed the 21 km course in 50:26 minute. Beat current human men's half-marathon world record of 57:20. Last year's winner took over 2 hours 40 minutes. Massive progress in 12 month

译在 2026 年北京人形机器人半程马拉松中，HONOR 的 Lightning 以 50 分 26 秒完成 21 公里赛程。 击败当前人类男子半程马拉松世界纪录 57 分 20 秒。 去年冠军耗时超过 2 小时 40 分钟。 12 个月内取得巨大进步

Rohan Paul@rohanpaul_ai · 4月19日

During halftime of a pro basketball game in Tokyo, Toyota's CUE7 robot rolled onto the court, stood from a seated position, dribbled the ball smoothly, and successfully did free throws in front of 8,400 fans. At 220 cm tall and 74 kg with a wheeled base

译在东京一场职业篮球赛的中场休息时，Toyota 的 CUE7 机器人滚动进入球场，从坐姿站起，流畅地运球，并在 8400 名观众面前成功完成罚球。 身高 220 厘米，重 74 公斤，带有轮式底座

Rohan Paul@rohanpaul_ai · 4月19日

This robot took home the “Best Design” award, in today's Beijing humanoid robot half-marathon. recognition that its motion looks closer to natural human running than most competitors. TienKung Ultra completed the full 21.1 km in 1 hour 15 minutes.

译这款机器人在今天的北京人形机器人半程马拉松中获得了"最佳设计"奖，以表彰其动作比大多数竞争对手更接近自然人类奔跑。 TienKung Ultra 在1小时15分钟内完成了全程21.1公里。

DogeDesigner@cb_doge · 4月19日

"Assuming the current trend of AI and robotics continues, which seems likely, the AI and robots will be able to do anything that that humans want them to do, essentially, so hopefully not more than that, but AI and robotics will be able to provide us all the goods and services that anyone could possibly want so you wouldn't need to work. People will be able to wherever they want with their free time. Work will be optional. I just want to separate out from like what I wish would happen versus what I predict will happen. Because people get confused about that. They think that what I predict will happen is what I wanted to happen. What I predict to happen is not the same as what I want to happen." 一 Elon Musk

译Elon Musk预测，若AI与机器人技术持续当前趋势，将能提供人类所需的一切商品与服务，物质产出将比现今高数个数量级，届时工作将变为可选项。他特别强调需区分"预测"与"愿望"，指出其关于未来丰裕社会的判断是基于技术趋势分析，而非个人主观期望，避免公众误解其立场。

DogeDesigner@cb_doge · 4月18日

Neuralink’s next product BlindSight aims to restore vision for people born blind. The implant has already shown success in monkeys. If this scales to humans, it’s not just a breakthrough. It’s millions of lives changed and an entirely new chapter for humanity.

译Neuralink 的下一款产品 BlindSight 旨在为先天失明者恢复视力。该植入物已在猴子身上取得成功。 如果这能推广到人类，这不仅仅是一项突破。它将改变数百万人的生命，并开启人类全新的篇章。

Chubby♨️@kimmonismus · 4月18日

You cant outrun Unitree H1. From jogging to running in seconds.

译你跑不过 Unitree H1。几秒内从慢跑到快跑。

Rohan Paul@rohanpaul_ai · 4月18日

The mystery is finally out. The robot behind that Times Square billboard is @XSquareRobot They officially posted "Yes — the Times Square billboard is ours!" They actually already launched China’s first robot-assisted home cleaning service in partnership with 58[.]com (a major household services platform, often called “China’s Craigslist”). Customers book via app; and a human cleaner and X Square robot (e.g., Quanta X1) arrive together. The robot handles repetitive tasks (wiping tables/surfaces, tidying, picking up debris) while the human manages judgment-heavy work. Real-home videos and reports from Shenzhen apartments confirm this is a paid, consumer-facing pilot service, not a lab demo. At roughly half the cost of a human-only cleaner—and early slots sold out quickly due to demand. Services have started in Shenzhen and Beijing, with more cities on the roadmap. Now they’re hosting a launch event in Beijing on April 21 to unveil their next embodied foundation model and talk more about their work in embodied intelligence.

译X Square 揭晓其时代广场广告中的机器人身份为 Quanta X1，并宣布已在中国推出首个机器人辅助家庭清洁服务。通过与 58.com 合作，用户可预约人类清洁工与机器人协同作业，机器人负责擦拭、整理等重复性劳动，人类处理复杂判断工作。该服务价格约为纯人工清洁的一半，目前已在深圳和北京启动试点，早期时段迅速售罄。公司将于4月21日发布新一代具身基础模型。

AK@_akhaliq · 4月17日46

HY-World 2.0 A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds paper: https://huggingface.co/papers/2604.14268

译HY-World 2.0 一个用于重建、生成和模拟3D世界的多模态世界模型 paper: https://huggingface.co/papers/2604.14268

DogeDesigner@cb_doge · 4月17日

"In the future, the robots will make so many robots, that they will actually saturate all human needs, meaning you won't be able to even think of something to ask the robot for at a certain point, like there will be such an abundance of goods and services. There'll be more robots than people. I think everyone on earth is going to have one humanoid robot because you would want a robot to watch over your kids, take care of your pet, take care of elderly parents. I'm very optimistic about the future. I think we're headed for a future of amazing abundance, which is very cool. Definitely we are in the most interesting time in history." 一 Elon Musk

译Elon Musk 描绘了一个由机器人驱动的极致富足未来。他认为未来机器人将实现自我复制并大规模普及，数量超过人类，达到每人拥有一台人形机器人的程度。这将彻底满足人类对商品和服务的所有需求，消除稀缺性。Musk 对前景表示极度乐观，称当前为"历史上最有趣的时期"，预示着一个物资极大丰富的时代即将到来。

DogeDesigner@cb_doge · 4月17日36

"For people that have lost a limb, arm or leg, In the future, we'll be able to attach an Optimus arm or legs, working with Neuralink and Tesla. So it goes far beyond just operating a robot hand, but replacing limbs and having kind of a whole body robot experience." 一 Elon Musk

译对于失去肢体、手臂或腿的人，未来我们将能够安装Optimus手臂或腿，与Neuralink和特斯拉合作。所以这远不止是操作机械手，而是替换肢体并获得一种全身机器人体验。

DogeDesigner@cb_doge · 4月17日

Elon Musk and Tesla are making roads safer and saving lives. Here’s a comparison of what Tesla haters’ propaganda shows vs the reality, where a Tesla FSD saved yet another kid’s life today by braking in time. Legacy Media will never show you this.

译埃隆·马斯克和 Tesla 正在让道路更安全并拯救生命。 这是 Tesla 仇恨者的宣传所展示的与现实的对比，现实中一辆 Tesla FSD 今天通过及时刹车再次救了一个孩子的命。 传统媒体永远不会给你看这个。

DogeDesigner@cb_doge · 4月17日

Tesla FSD saves a kid’s life. ❤️ A friend of mine in Arizona was on his way home in a Cybertruck using FSD when a kid suddenly ran across the road. The car reacted instantly and braked safely, avoiding what could’ve been a tragedy.

译Tesla FSD 救了一个孩子的命。❤️ 我的一位朋友在亚利桑那州，当时正开着 Cybertruck 用 FSD 回家，突然有个孩子跑过马路。车辆瞬间反应并安全刹车，避免了一场可能发生的悲剧。

AK@_akhaliq · 4月17日39

Geometric Context Transformer for Streaming 3D Reconstruction paper: https://huggingface.co/papers/2604.14141

译用于流式3D重建的几何上下文Transformer paper: https://huggingface.co/papers/2604.14141

Rohan Paul@rohanpaul_ai · 4月16日

Someone rented prime Times Square real estate to tease a new humanoid robot with 3 claims: - it has a "real brain," - works around the house, and - lets you "simply live." - "April 17th on X" Featuring a blurred humanoid alongside it! 🤔

译有人租下时代广场的黄金广告位来预告一款新的人形机器人，并提出了3个说法： - 它拥有"真正的大脑"， - 能做家务，并且 - 让你"简单生活"。 - "4月17日，在X上" 旁边还有一个模糊的人形机器人形象！🤔

AK@_akhaliq · 4月16日39

Habitat-GS A High-Fidelity Navigation Simulator with Dynamic Gaussian Splatting paper: https://huggingface.co/papers/2604.12626

译Habitat-GS 一种采用动态高斯泼溅的高保真导航模拟器 论文: https://huggingface.co/papers/2604.12626

DogeDesigner@cb_doge · 4月15日

Tesla just taped out the AI5 chip, huge milestone for FSD, Robotaxi & Optimus. • Single AI5 delivers ~5x the real-world compute of a dual AI4 setup • Massive leaps: ~8x compute power, 9x memory & 5x bandwidth vs current gen • One chip matches Nvidia H100 performance for Tesla workloads; dual setup rivals Blackwell, but at way lower cost & power draw • Radically simplified & optimized for edge AI inference (INT4/INT2/FP8 focus), perfect for cars & humanoid robots • Full AI5 computer targets 2,000–2,500 TOPS (vs ~300–500 for AI4) • “AI5 will make the cars almost perfect and greatly enhance Optimus” • “AI5 will punch far above its weight” thanks to Tesla’s tightly co-designed hardware + software stack Built in the USA (TSMC Arizona + Samsung Texas), with Terafab scaling up. Congratulations @elonmusk and @Tesla

译Tesla完成AI5芯片流片，单芯片算力达双AI4的5倍，整体算力提升8倍、内存9倍、带宽5倍。该芯片针对边缘AI推理深度优化，完整算力目标2000-2500 TOPS，单芯片性能匹敌Nvidia H100，双芯片可媲美Blackwell但成本功耗显著降低。AI5将大幅提升FSD、Robotaxi与Optimus性能，由TSMC Arizona与Samsung Texas在美国本土制造。

Rohan Paul@rohanpaul_ai · 4月15日

Some funny moments from Beijing’s 2026 Humanoid robot half-marathon

译北京 2026 人形机器人半程马拉松的一些有趣时刻 [引用 @rohanpaul_ai]：北京人形机器人半程马拉松，夜间测试。约 40% 的队伍完全自主运行，遥控跑获得 1.2× 罚分（4 月 19 日北京 E-Town 半程马拉松，100 多支队伍，300 多个人形机器人参赛）

Rohan Paul@rohanpaul_ai · 4月15日

Binghamton University demonstrated a robotic guide dog (Unitree Go2 base) that speaks naturally with users. In the test, it asked where the person wanted to go, suggested a route, then described surroundings in real time

译宾汉姆顿大学展示了一只机器导盲犬（Unitree Go2 底座），它能与用户自然对话。在测试中，它询问用户想去哪里，建议了一条路线，然后实时描述周围环境

AI Notkilleveryoneism Memes ⏸️@AISafetyMemes · 4月15日

Love how every AI safety concern from 5 years ago is just a Tuesday now

译喜欢现在5年前的每一个AI安全担忧都只是寻常周二了 [引用 @KaterynaLis]：‼️ 泽连斯基：战争史上首次，一个敌方阵地完全由地面机器人系统和无人机占领——没有任何步兵参与。机器人代替士兵进入了最危险的区域并占领了阵地。 「未来已至，就在战场上，而乌克兰正在创造它。这些是我们的地面机器人系统。在这场战争的历史上，首次完全由无人GRS平台和无人机占领敌方阵地。占领者投降了，这次行动没有步兵参与，我方也没有损失。Ratel、Termite、Ardal、Lynx、Zmiy、Protector、Volya和其他GRS在短短3个月内完成了超过22,000次前线任务。换句话说，超过22,000次生命被挽救。机器人代替士兵进入了最危险的区域」——泽连斯基对乌克兰国防工业综合体工作人员的讲话。2026年4月13日。

AK@_akhaliq · 4月14日47

Matrix-Game 3.0 Real-Time and Streaming Interactive World Model with Long-Horizon Memory paper: https://huggingface.co/papers/2604.08995

译Matrix-Game 3.0 具备长时记忆的实时流式交互世界模型 论文: https://huggingface.co/papers/2604.08995

Rohan Paul@rohanpaul_ai · 4月14日

Somewhere in Poland.. 😀 A Unitree G1 humanoid robot is on the trail of a pack of wild boars!

译在波兰某地.. 😀 一台 Unitree G1 人形机器人正在追踪一群野猪！

Chubby♨️@kimmonismus · 4月14日

Robots are coming. A Chinese tech company has unveiled a highly dexterous robotic hand capable of performing complex fine-motor tasks like finger games, solving Rubik’s cubes, and manipulating small objects.

译机器人来了。 一家中国科技公司发布了一款高度灵巧的机械手，能够执行复杂的精细动作任务，如手指游戏、解魔方和操作小物体。

AK@_akhaliq · 4月14日48

WildDet3D Scaling Promptable 3D Detection in the Wild paper: https://huggingface.co/papers/2604.08626

译WildDet3D 在野外扩展可提示的3D检测 论文: https://huggingface.co/papers/2604.08626

Rohan Paul@rohanpaul_ai · 4月13日

During Beijing’s recent humanoid robot half-marathon test runs, these bipedal robots demonstrated sustained running speeds that test their dynamic stability and energy management systems.

译在北京最近的人形机器人半程马拉松测试期间，这些双足机器人展示了持续的奔跑速度，考验其动态稳定性和能源管理系统。

Rohan Paul@rohanpaul_ai · 4月13日

Robotic data is insanely expensive and brutal to collect. It’s the only thing holding back general-purpose robots right now. Figure CEO @adcock_brett : "If we could get a pile of data in the helix stack, we would solve general robotics right now."

译机器人数据收集成本极高且过程艰难，这是当前阻碍通用机器人发展的唯一瓶颈。Figure CEO Brett Adcock表示，只要为helix stack提供足够数据，现在就能解决通用机器人问题。

Rohan Paul@rohanpaul_ai · 4月13日

India is quietly becoming a training floor for humanoid robots, with workers filming thousands of first-person hand tasks so AI systems can learn grasping, folding, sorting, and tool use. This story is really about how the humanoid robot boom still depends on cheap, repetitive human labor to teach machines basic physical skill. The problem is that robots do not fail on big plans first; they fail on tiny physical details like grip angle, finger timing, slip correction, and object contact. That kind of knowledge is hard to code and expensive to collect. These labs capture that missing layer by putting cameras or sensors on people and recording ordinary actions as machine-readable motion examples. The useful part is not the towel or box itself but the sequence: where the hand starts, how force changes, when fingers adjust, and how the body recovers from small mistakes. That gives robotics teams supervised data for models that map visual input to physical actions, which is much easier than hand-coding every movement rule. This is a story about how physical intelligence gets extracted before it gets automated. --- quasa. io/media/the-hidden-hand-farms-of-india-fueling-the-ai-robot-revolution-with-human-motion

译印度工人佩戴头戴摄像头采集第一人称手部动作数据，用于训练人形机器人掌握抓取、折叠等物理技能。这揭示了当前机器人热潮仍依赖廉价人类劳动获取 embodied data。与语言模型不同，机器人需从人类动作中学习握持角度、力度调整等微观物理细节。该模式将人类劳动双重商品化：既是生产工作，又成为训练AI的数据基础设施。在具身数据采集成本降低前，机器人行业将持续依赖工人劳动作为"物理智能"的廉价来源。

Rohan Paul@rohanpaul_ai · 4月12日

Unitree’s H1 humanoid reached 10 m/s. That is world-champion territory, Usain Bolt's peak was 12.42 m/s, with his 100 m average around 10.44 m/s. Unitree was 3.3 m/s just 2 years ago. significant advances in balance & gait control at high velocities

译Unitree H1 人形机器人奔跑速度达到 10 m/s，接近博尔特百米平均速度（10.44 m/s）。较两年前的 3.3 m/s 大幅提升，标志着高速平衡与步态控制技术取得关键突破。

Chubby♨️@kimmonismus · 4月11日

Unitree makes a humanoid that runs at 10m/s (Bolt runs at 12.42 m/s) Ngl pretty impressive

译Unitree 人形机器人奔跑速度达到 10m/s，已接近 Bolt 保持的 12.42 m/s 人类短跑纪录。不得不说，这一速度表现相当 impressive。

AK@_akhaliq · 4月11日

HY-Embodied-0.5 Embodied Foundation Models for Real-World Agents paper: https://huggingface.co/papers/2604.07430

译HY-Embodied-0.5正式发布，专为真实世界智能体打造的具身基础模型，相关论文已公开至Hugging Face。

SemiAnalysis@SemiAnalysis_ · 4月8日

Unitree filing that their blended 59% gross margins in 2025 YTD is incredible! But SemiAnalysis newsletter subscribers knew this nearly ~6 months ago. Here is our BoM from behind the paywall for proof😉 If you're wondering how 59% was possible, you're probably also missing why quadrupeds are already viable employees (1/2)🧵

译Unitree 披露其 2025 年至今混合毛利率达 59%，这太不可思议了！

AK@_akhaliq · 4月7日

OpenWorldLib A Unified Codebase and Definition of Advanced World Models paper: https://huggingface.co/papers/2604.04707

译OpenWorldLib 正式发布，提供高级世界模型的统一代码库与标准化定义，相关论文已上传至 Hugging Face。

Jim Fan@DrJimFan · 4月1日

The power of the Claw, in the palm of a robot hand. Agentic robotics is here! Today, we open-source CaP-X: vibe agents, alive in the physical world. They incarnate as robot arms and humanoids with a rich set of perception APIs, actuation APIs, and auto synthesize skill libraries as they go. CaP-X is a strict superset of our old stack, because policies like VLAs are “just” API calls as well. It solves many tasks zero-shot that a learned policy would struggle with. And we are doing much more than vibing. CaP-X is our most systematic, scientific study on agentic robotics so far: - We build a comprehensive agentic toolkit: perception (SAM3 segmentation, Molmo pointing, depth, point cloud), control (IK solvers, grasp planner, navigation), and visualization (EEF, mask overlays) that work across different robots. - CaP-Gym: LLM’s first Physical Exam! 187 manipulation tasks across RoboSuite, LIBERO-PRO, and BEHAVIOR. Tabletop, bimanual, mobile manipulation. Sim and real. Can’t wait to see the gradients flow from CaP-Gym to the next wave of frontier LLM releases. - CaP-Bench: we benchmark 12 frontier LLMs/VLMs (Gemini, GPT, Opus, Qwen, DeepSeek, Kimi, and more) across 8 evaluation tiers. We systematically vary API abstraction level, agentic harness, and visual grounding methods. Lots of insights in our paper. - CaP-Agent0: a training-free agentic harness that matches or exceeds human expert code on 4 out of 7 tasks without task-specific tuning. - CaP-RL: if you get a gym, you get RL ;). A 7B OSS model jumps from 20% to 72% success after only 50 training iterations. The synthesized programs transfer to real robots with minimal sim-to-real gap. 3 years ago, our team created Voyager, one of the earliest agentic AI that plays and learns in Minecraft continuously. Its key ideas — skill libraries, self-reflection loops, and in-context planning — have since influenced many modern agentic designs. Today, the agent graduates from Minecraft and gets a real job. It’s April Fool’s, but this Claw is getting its hands dirty for real! Link in thread:

译CaP-X开源具身智能系统，让大模型智能体通过机械臂与人形机器人进入物理世界。系统整合SAM3、Molmo等感知API与IK求解器、抓取规划等控制接口，可自动合成技能库。研究发布CaP-Gym基准（187项操作任务）与CaP-Bench（评测12个前沿模型），提出零样本框架CaP-Agent0及强化学习方案CaP-RL，后者仅用50次迭代即将7B模型成功率从20%提升至72%。该技术由曾开发Minecraft智能体Voyager的团队推出。

Demis Hassabis@demishassabis · 3月25日

Excited to partner with Agile Robots! Looking forward to seeing our models being deployed through Agile Robots incredible platform to help solve some of the most complex industrial challenges

译Google DeepMind 宣布与 Agile Robots 建立研究合作，将 Gemini 基础模型集成至对方硬件平台，部署于工业场景解决复杂挑战，构建下一代更实用的机器人。

Google DeepMind@GoogleDeepMind · 3月24日

Google DeepMind 🤝 Agile Robots Our new research partnership will integrate the Gemini foundation models with their hardware to help build the next generation of more helpful and useful robots. Find out more → https://goo.gle/4lKu7de

译Google DeepMind 与 Agile Robots 建立研究合作，将 Gemini 基础模型与后者硬件集成，共同开发下一代更实用的机器人。

Saining Xie@sainingxie · 3月24日

best read paired with the LeWorldModel paper. don’t ask me why 🙂

译最好搭配 LeWorldModel 论文阅读。别问我为什么 🙂

Jim Fan@DrJimFan · 3月24日

Teleop is so 2025. Ever since we unveiled EgoScale and the dexterity scaling law, it's been clear to us and the ecosystem that behavior cloning directly from humans is the way to break the curse of teleop. 2026 is all about scaling robot learning without robots.

译EgoVerse 生态系统正式发布：由4家研究实验室与3家行业伙伴共建，基于1300+小时第一人称人类视频数据，覆盖240个场景与2000+项任务。无需真实机器人即可规模化训练，通过行为克隆直接从人类数据学习，2026年机器人学习将彻底告别遥操作模式。

Andrej Karpathy@karpathy · 3月19日

Thank you Jensen and NVIDIA! She’s a real beauty! I was told I’d be getting a secret gift, with a hint that it requires 20 amps. (So I knew it had to be good). She’ll make for a beautiful, spacious home for my Dobby the House Elf claw, among lots of other tinkering, thank you!!

译Andrej Karpathy 收到首台 DGX Station GB300（Dell Pro Max with GB300），这台需要 20 安培电流的"秘密礼物"将成为 Dobby the House Elf claw 等项目的宽敞新家。

Yann LeCun@ylecun · 3月10日

Unveiling our new startup Advanced Machine Intelligence (AMI Labs). We just completed our seed round: $1.03B / 890M€, one the largest seeds ever, probably the largest for a European company. We're hiring! [the background image is the Veil Nebula - a picture I took from my backyard, most appropriate for an unveiling] More details here: https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/

译AMI Labs 宣布获 10.3 亿美元种子轮融资，由 Cathay Innovation、Bezos Expeditions 等领投，创欧洲公司种子轮纪录。公司致力于开发基于世界模型的新型 AI 系统，具备理解世界、持久记忆、推理规划及可控安全能力，团队分布于巴黎、纽约、蒙特利尔和新加坡。

Saining Xie@sainingxie · 3月10日

i’m joining forces with @ylecun and an incredible group of people to start AMI Labs @amilabs. AMI isn’t a conventional lab. we don’t intend to become one. a lot to say about why this moment matters, but for now we’re heads down building. join us: https://amilabs.xyz

译我与 @ylecun 和一群杰出人士联手创立 AMI Labs @amilabs。

Saining Xie@sainingxie · 2月27日

world modeling is never about rendering pixels. rendering is local. world state is global. as soon as more than one agent exists, the only thing that truly matters is the shared representation beneath individual views. that shared representation is what scales into collective capability. this is why I'm super excited to share project Solaris -- our new work focused on building a multiplayer video world model in minecraft. This release includes three main pieces. 1⃣Solaris Engine, a fully featured multiplayer data collection system with built in visuals. the team put a huge amount of work into this since nothing like it really exists yet. https://github.com/solaris-wm/solaris-engine 2⃣Solaris Model, a multiplayer DiT with a new memory efficient self forcing design, trained on 12.6M frames of coordinated Minecraft gameplay. https://github.com/solaris-wm/solaris 3⃣Solaris Eval, which uses a VLM as a judge to evaluate different multiplayer capabilities. read the full technical breakdown by @ojmichel4, and start building with Solaris. https://solaris-wm.github.io/

译Project Solaris提出世界建模的本质在于全局共享状态而非局部像素渲染，推出基于Minecraft的多人在线视频世界模型。该系统突破单智能体视角局限，支持任意数量智能体随时介入交互，实现持久化世界状态演化。核心包含三大组件：Solaris Engine多人数据收集系统、基于DiT架构的Solaris Model（采用新型内存高效自强制设计，训练于1260万帧协调游戏数据）、以及使用VLM评判的Solaris Eval评估体系。这一范式转变为构建神经MMORPG服务器奠定基础。