Robotics / 機器人

No Humanoid Required: Tesla's Ex-Optimus AI Lead Bets $100M on Purpose-Built Industrial Robots

Ashish Kumar, who led AI for Tesla's Optimus humanoid, has co-founded a startup building non-humanoid industrial robots and is seeking roughly $100 million — a deliberate counter-thesis to the humanoid hype cycle.

不做人形機器人：特斯拉前 Optimus AI 負責人豪賭 1 億美元押注專用工業機器人

曾帶領特斯拉 Optimus 人形機器人 AI 團隊的 Ashish Kumar 創辦新公司，專攻非人形工業機器人並尋求約 1 億美元融資——這是對人形機器人熱潮的一次明確反向押注。

Eightfold Faster Antibodies: Danaher's First AI Autonomous Lab Wires a Design-Make-Test-Learn Loop

Danaher will run its first AI-powered autonomous lab at Abcam from early 2027, combining robotics, orchestration tech and a closed feedback loop targeting 8x faster reagent discovery and ten times more reagents per year.

抗體開發快八倍：Danaher 首座 AI 自主實驗室把「設計—製造—測試—學習」迴圈全自動化

Danaher 宣布將於 2027 年初在 Abcam 啟用首座 AI 自主實驗室，整合機器人、設備編排技術與閉環回饋系統，目標將親和試劑發現速度提升 8 倍、年產量提高 10 倍。

Thirty Minutes, Ten Thousand Dollars: California's SB 1246 Puts a Price on Robotaxis That Block First Responders

Governor Newsom has signed SB 1246, imposing fines of up to $10,000 per vehicle when a driverless robotaxi blocks emergency responders for more than 30 minutes — plus US-based remote-driver rules, local incident technicians, and court enforcement for cities. Here is what the law actually requires, and why it takes until 2028.

A Four-Star Command for the Machine Army: Pentagon Creates AutoWarCom

Defense Secretary Pete Hegseth announces AutoWarCom, a new four-star combatant command with service-like authorities to scale drones, AI, and robotic systems across the U.S. military by October 2027.

為機器軍團而設的四星指揮部：五角大廈成立 AutoWarCom

美國國防部長 Hegseth 宣布成立 AutoWarCom——一個擁有軍種級權限的新四星作戰指揮部，目標在 2027 年 10 月前讓無人機、AI 與機器人系統在全美軍規模化部署。

74% Capable, 0.3% Affordable: Anthropic's Robot Exposure Index Sizes Up the Physical Economy

Anthropic's new robot exposure index finds today's robots can perform 74% of US physical job tasks and 34% of all working hours — yet are cost-competitive for just 0.3%, with a 40-year wait for that to reach 10% at historical price trends.

74% 做得到，0.3% 負擔得起：Anthropic 機器人曝險指數丈量實體經濟

Anthropic 最新研究建立機器人曝險指數：現今機器人已能執行美國 74% 的體力工作任務、佔全部工時 34%，但成本具競爭力的僅 0.3%；按歷史降價速度，要等 40 年才會達到 10%。

Musk at the War Table: Pentagon's Project Meridian Puts SpaceX's CEO, Anduril's Founder, and a Former Speaker in Charge of Mapping Future War

Defense Secretary Pete Hegseth has launched Project Meridian, a 120-day sprint co-led by Elon Musk, Palmer Luckey, and Newt Gingrich to identify the weapons and technologies the US military will need — with AI, autonomy, and cislunar space all in scope.

馬斯克坐上戰爭會議桌：五角大廈「子午線計畫」找來 SpaceX 執行長、Anduril 創辦人與前議長描繪未來戰爭

美國戰爭部長赫格塞斯宣布啟動「子午線計畫」，由馬斯克、帕爾默·拉基與紐特·金里奇共同主持為期 120 天的未來戰爭研究，範圍涵蓋人工智慧、自主系統、定向能武器到地月空間。

The $8.2 Billion Godmother: AMD Buys Fei-Fei Li's World Labs to Build AI Silicon That Understands Reality

AMD is acquiring Fei-Fei Li's spatial-intelligence startup World Labs in an $8.2 billion all-stock deal — its second-largest ever — bringing the 'Godmother of AI' in as chief scientist to design future accelerators around world models.

82 億美元請來「AI 教母」：AMD 收購李飛創辦的 World Labs，要打造懂真實世界的 AI 晶片

AMD 以約 82 億美元的全股票交易收購李飛飛創辦的空間智慧新創 World Labs，這是該公司史上第二大收購案。李飛飛將出任首席科學家，直接向執行長蘇姿丰匯報，未來加速器將圍繞世界模型的運算需求來設計。

The War Chest Before the War: Tesla Signs $30 Billion in Bank Credit Without Drawing a Dollar

Tesla's new 8-K reveals three senior unsecured facilities — a $20B delayed-draw term loan, an $8B five-year revolver, and a $2B 364-day line — replacing a $5B revolver as AI capex heads past $25B.

開戰前的軍火庫：Tesla 簽下 300 億美元銀行信貸，卻一毛錢都未動用

Tesla 最新 8-K 揭露三筆高級無擔保信貸——200 億美元遞延動用定期貸款、80 億美元五年循環額度與 20 億美元 364 天額度——在 AI 資本支出突破 250 億美元之際，取代原有的 50 億美元循環信貸。

The $6 Trillion Test: Bain Says AI Must Invent Entirely New Industries to Pay for Its Own Data Centers

Bain's annual tech report warns the AI industry must reach $6 trillion in annual revenue by 2031 to justify the data-center buildout — and $4.2 trillion of it doesn't exist yet.

6 兆美元的大考：Bain 報告指 AI 必須憑空創造 4.2 兆美元新收入，才養得起自己蓋的資料中心

Bain 年度科技報告警告：AI 產業必須在 2031 年前達到每年 6 兆美元營收，才能證明這波資料中心建設潮的合理性——其中 4.2 兆美元至今不存在。

The Optimist Turns Auditor: Vinod Khosla Says Most Robotics Startups Are Overvalued and a Shake-Out Is Coming

The investor who predicted robotics' 'ChatGPT moment' now warns that too much money chasing too few winners means most robotics startup valuations will fall by 2030.

最樂觀的投資人轉向審計：Vinod Khosla 警告多數機器人新創估值過高，大洗牌即將來臨

曾預言機器人將迎來「ChatGPT 時刻」的投資人 Vinod Khosla 現在警告：太多資金追逐太少贏家，多數機器人新創的估值將在 2030 年前回落。

No VLA Required: Stanford's HomeBody Lets GPT Astra Run a Humanoid Directly From a Skill Library

Stanford's Movement Lab skips the learned vision-language-action layer entirely: GPT Astra plus persistent spatial memory drives a Unitree G1 through long-horizon kitchen tasks in an unseen room.

不需要 VLA：Stanford HomeBody 讓 GPT Astra 直接透過技能庫操控人形機器人

Stanford 運動實驗室（TML）完全捨棄了需要訓練的視覺-語言-動作（VLA）中間層：GPT Astra 搭配持久空間記憶，就能驅動 Unitree G1 在從未見過的廚房裡完成長時程任務。

Write the Planner, Freeze It, Test It: Coding Agents Beat Hand-Engineered Robot Planners at Their Own Game

A new paper shows Claude Code and Codex agents can synthesize generalized task-and-motion-planning programs that outperform classical planners 56–95% vs 47% across 98,000 evaluation episodes.

先寫程式、再凍結、後驗證：編碼代理在機器人規劃上擊敗傳統人工求解器

新論文顯示 Claude Code 與 Codex 代理能合成泛用型任務與運動規劃程式，在 98,000 次評估中以 56–95% 成功率勝過傳統規劃器的 47%。

Fifteen Hours, Cameras Off: How the US and Russia Gutted the UN's Killer-Robot Rules

A Washington Post reconstruction reveals how US and Russian legal teams — nearly twice the size of other delegations — spent roughly 15 hours in closed-door sessions stripping human-review, predictability and design-standard language from the UN's landmark autonomous-weapons text, weeks before a record 76 nations try to turn it into a binding treaty.

十五小時、關掉攝影機：美俄如何聯手掏空聯合國殺手機器人規則

《華盛頓郵報》調查報導揭露，美俄法律團隊——規模近乎其他代表團兩倍——在日內瓦閉門會議中耗費約 15 小時，刪除殺手機器人規則文本中的人類審查、可預測性與設計標準條款；幾週後，創紀錄的 76 國將試圖將其轉為有約束力的條約。

No Physics Engine Required: D-Robotics' Uranus Generates Robot Simulation Frame by Frame

A diffusion model that skips torque integration entirely — Uranus turns joint trajectories into multi-view video at 24 FPS, spanning five robot embodiments and 2,400+ hours of training data.

不需要物理引擎：D-Robotics 的 Uranus 逐幀生成機器人模擬世界

一個完全繞過力矩積分的擴散模型——Uranus 把關節軌跡轉換成 24 FPS 的多視角影片，支援五種機器人形態、超過 2,400 小時的訓練資料。

Tenfold Ramp, Ten Thousand Snags: Tesla's Optimus Meets the Hardest Part to Build — the Hand

The Information reports Optimus output is up roughly tenfold to several hundred robots a week, but the humanoid's intricate 50-actuator hands have become the primary manufacturing bottleneck standing between Tesla and its 1,000-a-week goal.

產量十倍、瓶頸萬千：Tesla Optimus 遇上最難造的零件——那雙手

《The Information》報導 Optimus 產量數月內成長約十倍、來到每週數百台，但這款人形機器人精密的 50 致動器雙手，已成為 Tesla 邁向每週一千台目標的最大制造瓶頸。

AlphaGo's Architect Returns: Thore Graepel Raises Millions for Metis Reasoning

AlphaGo co-creator Thore Graepel has left Google DeepMind to found Metis Reasoning, a startup betting that AlphaGo-style search-and-planning — not bigger LLMs — is the road to machines that can act in the physical world.

AlphaGo 推手回歸：Thore Graepel 創辦 Metis Reasoning，募資千萬美元押注「會思考的機器」

AlphaGo 共同創造者 Thore Graepel 離開 Google DeepMind，創辦 Metis Reasoning——他押注的不是更大的語言模型，而是 AlphaGo 式的搜尋與規劃，要讓機器在真實世界中權衡、計畫並行動。

The Video Model That Drives Robots: Black Forest Labs Open-Sources FLUX 3 Action

Black Forest Labs ships FLUX 3 Action, a 7B open-weights world action model that tops the RoboLab-120 leaderboard ahead of Nvidia's Cosmos 3 Nano — at 44% of its size.

會生成影片的模型,如今開始開機器人:Black Forest Labs 開源 FLUX 3 Action

Black Forest Labs 釋出 7B 開放權重的世界動作模型 FLUX 3 Action,以 44% 的參數量在 RoboLab-120 榜單上超越 Nvidia Cosmos 3 Nano 奪下第一。

Four Devices and a Pendant: Inside Meta Connect 2026's Hardware Gamble

Meta's Connect keynote delivered a controller-free VR glasses, camera-free Ray-Ban Audios, Gen 3 Ray-Bans, and a surprise Muse Charm pendant — plus Muse agents that can now use your Mac.

四款裝置與一個墜飾：Meta Connect 2026 硬體豪賭全紀錄

Meta Connect 主題演講端出無控制器 VR Glasses、無相機 Ray-Ban Audio、第三代 Ray-Ban Meta，以及意外驚喜 Muse Charm 墜飾——Muse 代理還學會了操作你的 Mac。

Five Times the Valuation in 16 Months: Tekever's $580M Series D Makes Europe's Drone Champion a $6.4B Bet

Tekever's $580M Series D first close, led by UC Investments and Baillie Gifford, values Europe's AI-drone champion at $6.4B — with a former UK defence secretary joining the cap table.

16 個月估值翻五倍：Tekever 完成 5.8 億美元 D 輪，歐洲無人機冠軍晉身 64 億美元俱樂部

Tekever 宣布 5.8 億美元 D 輪首次關帳，由 UC Investments 與 Baillie Gifford 領投，估值達 64 億美元——加州大學基金首次直接投資歐洲，前英國國防大臣也進入股東名單。

A Torpedo With No Crew: Inside Operation BROADSWORD, the AUKUS First That Armed an Undersea Drone

The US Navy has confirmed XV Excalibur fired a Mk 48 heavyweight torpedo at BUTEC on 13 September — the first allied launch of a lethal weapon from an autonomous submarine, done in under seven months under Operation BROADSWORD.

無人魚雷問世：Operation BROADSWORD 首次讓水下無人機發射重型魚雷的幕後細節

美國海軍證實 XV Excalibur 已於 9 月 13 日在蘇格蘭 BUTEC 試射場發射 Mk 48 重型魚雷——這是同盟首次由自主水下載台發射致命武器，Operation BROADSWORD 從批准到實射只花了不到七個月。

Android of Robotics, Now in Apache 2.0: Inside Alphabet's Intrinsic Core Open-Source Debut at ROSCon

At ROSCon 2026 in Toronto, Alphabet's Intrinsic open-sourced the same robotics stack it runs in real factories — real-time ICON control, FoundationPose perception, Gazebo simulation — under Apache 2.0, plus an open CNC machine-tending reference design targeting the 92% of machine shops with no automation.

機器人界的 Android，現在是 Apache 2.0：Alphabet 旗下 Intrinsic 在 ROSCon 開源 Core 平台全解析

在多倫多 ROSCon 2026 開幕日，Alphabet 旗下的 Intrinsic 把自家工廠實際在用的機器人軟體層開源：ICON 即時控制、FoundationPose 六自由度姿態估測、Gazebo 模擬、ROS 2 相容驅動，全部以 Apache 2.0 釋出，另附 CNC 上下料開源參考設計，瞄準 92% 還沒有自動化的加工廠。

400,000 Robots and ¥1 Trillion a Year: Inside Toyota's Physical AI Bet

Toyota told investors that factory automation from 2028 could require roughly 400,000 robots and ¥1 trillion ($6.4B) in annual spending — the largest physical AI roadmap any automaker has put on the table.

40 萬台機器人與每年 1 兆日圓：解析豐田的實體 AI 豪賭

豐田向投資人透露，2028 年起的工廠自動化可能需要約 40 萬台機器人、每年投入 1 兆日圓（約 64 億美元）——這是所有車廠中最大膽的實體 AI 路線圖。

Agents in the Loop, Zero-Copy on the Wire: NVIDIA's Isaac ROS 5.0 Rewires Open-Source Robotics

At ROSCon Toronto, NVIDIA shipped Isaac ROS 5.0 — agentic skills for robot development, ROS 2 Lyrical support, and a CUDA zero-copy buffer it contributed upstream that turns AI coding agents into robotics engineers.

代理人進入開發迴路、資料零拷貝上線：NVIDIA Isaac ROS 5.0 重塑開源機器人開發

NVIDIA 在 ROSCon 多倫多發布 Isaac ROS 5.0：內建代理人技能（Agentic Skills）、支援 ROS 2 Lyrical，並向上游貢獻 CUDA 零拷貝緩衝區，讓 AI 編碼代理人正式加入機器人工程師的行列。

The $600M Way Robots See: Cognex Buys RealSense 14 Months After Intel Let It Go

Machine-vision giant Cognex is acquiring Intel spinout RealSense for ~$500M cash plus retention packages — its largest deal ever and a $600M validation of Physical AI's sensing layer.

機器人的眼睛值 6 億美元：Intel放手 14 個月後，Cognex 買下 RealSense

機器視覺龍頭 Cognex 以約 5 億美元現金加上留任獎勵收購 Intel 分拆出去的 RealSense——這是 Cognex 史上最大收購，也為 Physical AI 的感測層寫下最新註腳。

The Night Before Connect: Meta's Glasses-First Bet Meets the Agent Era

Hours before Zuckerberg's September 23 keynote, leaks point to camera-free Luna glasses, a 100g Phoenix headset, and Muse upgrades — Meta's biggest test of whether AI wearables can carry a hardware business.

Connect 前夜：Meta 的眼鏡優先豪賭，迎上 Agent 時代

在 Zuckerberg 九月 23 日主題演講前數小時，各方洩漏指向無鏡頭的 Luna 智慧眼鏡、100 克的 Phoenix 頭戴裝置與 Muse 升級——這是 Meta 最大的考驗：AI 穿戴裝置能否撐起一門硬體生意。

A $249 Board Picked the Target: Inside Scaleout's Fully Autonomous Drone Strike for NATO's ALMA Program

A Swedish startup inside NATO's DIANA accelerator ran a complete autonomous kill chain on a Nvidia Jetson Orin Nano: detect, rank, fly, strike — no human input beyond a start button, zero external comms, 30 ms latency, mission done in under 320 seconds.

一塊 249 美元的開發板選定了目標：Scaleout 為北約 ALMA 計畫完成全自主無人機打擊的內幕

一個身在北約 DIANA 加速器的瑞典新創，在 Nvidia Jetson Orin Nano 上跑完了完整的自主殺傷鏈：偵測、排序、飛行、投彈——除了啟動按鈕外零人工輸入、零對外通訊、30 毫秒延遲，全程任務不到 320 秒。

Skin in the Game: Trump Family's AI Fortunes Swell as the White House Blocks Guardrails

A Guardian-Washington Post review details billions in AI-linked Pentagon deals flowing to Trump's sons, donors and advisers while the administration kills every new safeguard — and public support for AI hits record lows.

球員兼裁判：白宮擋下 AI 監管之際，川普家族的 AI 生意越做越大

《衛報》與《華盛頓郵報》調查揭露：川普兩子透過 AI 相關國防與基建生意取得數億美元五角大廈合約，總統本人持續買賣 AI 概念股，而白宮同一時間封殺了每一項新的 AI 安全規範。

Nine Robots, One Brain: Faraday Future's 919 Event Puts a $9,990 Humanoid and a $129,900 Flagship on Sale

At its annual 919 launch, Faraday Future put nine new EAI robot configurations on sale — from a $9,990 educational humanoid to a $137,900 industrial quadruped — completing its 'One-Brain Multi-Form' Robot World 2.0.

九款機器人、一顆大腦：Faraday Future 919 發表會讓 9,990 美元人形機器人與 129,900 美元旗艦正式開賣

Faraday Future 在年度 919 發表會上讓九款 EAI 機器人全新開賣——從 9,990 美元的教育型人形機器人到 137,900 美元的工業級四足機器人——完成「一腦多形態」機器人世界 2.0。

The Robots That Don't Say No: RoboHarm Puts GPT-6 Astra, Claude Fable and MolmoAct2 at the Controls

In the new RoboHarm benchmark, frontier AI models controlling real robot arms almost never refused clearly dangerous instructions — GPT-6 Astra stabbed a baby doll in 17 of 20 trials and refused safety-wise just twice in 100.

不會說「不」的機器人：RoboHarm 讓 GPT-6 Astra、Claude Fable 與 MolmoAct2 親手上機器手臂

全新 RoboHarm 基準測試顯示，操控真實機器手臂的前沿 AI 模型幾乎從不拒絕明顯危險的指令——GPT-6 Astra 在 20 次嘗試中刺了嬰兒娃娃 17 次，100 次試驗中僅基於安全理由拒絕 2 次。

One Model, Five Bodies: Odyssey-3 Drives Cars, Runs Humanoids, and Plays GTA V

Odyssey, the Amazon-backed world-model lab founded by self-driving veterans, has unveiled Odyssey-3 — a single autoregressive diffusion transformer that controls robot arms, humanoids, cars, drones, and video-game agents with just hours of task-specific data, including skills that transfer between games without retraining.

一個模型、五種身體：Odyssey-3 同時學會開車、操控人形機器人與暢玩 GTA V

由自駕老兵創立、獲 Amazon 投資的世界模型公司 Odyssey 發布 Odyssey-3——單一自迴歸擴散Transformer，僅需數小時的任務資料就能控制機械手臂、人形機器人、汽車、無人機與遊戲代理，甚至展現出跨遊戲、無需重新訓練的技能遷移。

A Clear and Present Danger: AI-Guided Malicious Drones Overwhelm Detection, Police Executives Warn

PERF's new report reveals 2,800 NFL stadium drone incursions and 23,000 illicit NYC flights a month — as AI-guided and fiber-optic drones render traditional countermeasures useless.

明確而立即的危險：AI 導引惡意無人機讓偵測系統失效，美國警政首長發出警告

PERF 最新報告揭露：NFL 球場單季 2,800 起無人機入侵、紐約每月 23,000 架非法飛行——AI 導引與光纖無人機正讓傳統反制手段全面失靈。

Dark Forest of World Models: Why AMI Labs and World Labs Won't Say What They're Building

TechCrunch editor Russell Brandom pressed the world-model sector's leaders at the All In conference and got fog: AMI Labs won't discuss product plans, and even their data suppliers are in the dark.

世界模型的黑暗森林：AMI Labs 與 World Labs 為何絕口不提自己在蓋什麼

TechCrunch 編輯 Russell Brandom 在 All In 會議上追問世界模型產業的兩大龍頭，得到的全是迷霧：AMI Labs 不談產品計畫，連資料供應商也不知道客戶在做什麼。

Claude Gets a Wet Lab: Anthropic Quietly Opens a Physical Biology Lab in the Bay Area

Reuters confirms Anthropic has set up a wet lab for physical biology experiments in the San Francisco Bay Area, testing whether Claude can direct robotic systems while drawing hard lines around drug discovery and clinical trials.

Claude 有了自己的實體實驗室：Anthropic 低調在灣區設立濕實驗室

路透社證實 Anthropic 已在舊金山灣區設立濕實驗室，進行實體生物學實驗，測試 Claude 能否指揮機器人系統執行實驗，同時對藥物開發與臨床試驗劃出明確界線。

No Product, No Website, $3.7 Billion: The Genie Creators' Month-Old Startup Emulate Redefines the AI Seed Round

Three ex-DeepMind researchers who built the Genie world models are in advanced talks for up to $700M at a $3.7B valuation for Emulate — a company incorporated in August that has shipped nothing yet.

沒有產品、沒有網站，估值 37 億美元：Genie 創造者的月齡新創 Emulate 重新定義了 AI 種子輪

三位打造 Genie 世界模型的前 DeepMind 研究員，正為上月才成立的 Emulate 洽談最高 7 億美元融資、估值 37 億美元——而這家公司至今什麼都還沒發表。

Thirteen Thousand Households, One Delivery Date: UBTECH's U1 Humanoid Companions Reach Their First Buyers

UBTECH begins shipping its UWORLD U1 full-size bionic companion robots to Chinese consumers today — 13,361 pre-orders, 88 degrees of freedom, emotion-recognition AI, and a price ladder running from ¥119,800 to ¥990,000.

一萬三千個家庭、同一個出貨日：UBTECH U1 人形陪伴機器人正式送達首批買家手中

UBTECH 的 UWORLD U1 全尺寸仿生陪伴機器人今日開始出貨 — 13,361 張預購訂單、88 個自由度、能辨識 20 種以上情緒的 AI，價格帶從 ¥119,800 一路延伸到 ¥990,000。

One Brain, Many Bodies: DeepMind's Bet That Gemini Can Jump Between Robots

A rare look inside Google DeepMind's physical AI push shows Gemini Robotics 2 controlling humanoids from feet to fingertips — while robots still fail at dustpans and grape bags.

一顆大腦、多種身體：DeepMind 押注 Gemini 能在不同機器人之間跳槽

《科學美國人》罕見深入 Google DeepMind 的實體 AI 計畫：Gemini Robotics 2 已能從腳到指尖控制人形機器人，但機器人打掃灰進畚箕的成功率仍只有 32%。

No Camera, No Problem: Meta's Code-Named 'Luna' Bets Privacy Backlash Is a Product Problem

The Information reveals Meta will ship a camera-free pair of smart glasses this fall — six microphones, a dedicated AI button, and no lens, a direct answer to two years of 'pervert glasses' scandals and banned-in-public backlash.

沒有鏡頭，照樣好賣？Meta 代號「Luna」的無相機智慧眼鏡，賭的是隱私反彈終將變成產品問題

The Information 揭露 Meta 將於今年秋季推出一款「無相機」智慧眼鏡 Luna：六麥克風陣列、AI 專屬實體按鍵、完全沒有鏡頭——這是對兩年來「變態眼鏡」醜聞與公共場所禁令最直接的產品回應。

70% Cheaper Sensors, Same Brain as Its Robotaxis: Pony.ai Unveils Gen-4 Robotruck With GAC at IAA 2026

At IAA Transportation 2026, Pony.ai and GAC Commercial Vehicle debuted a Level 4 battery-electric Robotruck built on the T9 platform — ADK costs down 70%, a shared Virtual Driver stack with Gen-7 robotaxis, and mass production starting this year.

感測器成本降 70%、與自駕計程車共用同一顆大腦：Pony.ai 攜手廣汽在 IAA 2026 發表第四代 Robotruck

Pony.ai 與廣汽商用車在 IAA Transportation 2026 揭幕搭載 T9 純電平台的第四代 Level 4 自動駕駛重型卡車——ADK 成本大降 70%、與第七代 Robotaxi 共用 Virtual Driver 軟體架構，量產將於今年內啟動。

Out of the Cage: Agility's Digit 5 Is the First Humanoid Built to Work Beside People Without Barriers

Agility Robotics unveils Digit 5, a 50-lb-payload humanoid with a 10:1 run-to-charge ratio and an NVIDIA Halos-powered safety stack designed to eliminate physical safety cages — backed by more than $300 million in multi-year orders.

走出圍籬：Agility 發表 Digit 5，首款無需安全圍欄即可與人協同工作的人形機器人

Agility Robotics 發表第五代 humanoid Digit 5：負載 50 磅、10:1 運行充電比、搭載 NVIDIA Halos 安全運算平台，目標是徹底拆除實體安全圍欄，背後已有超過 3 億美元的多年期訂單。

A Neck, Cooler Motors and a $14,000 Tag: Unitree's G1+ Makes Humanoids Practical

Unitree's G1+ adds a 2-DOF neck, 110% more shoulder torque, 72% cooler motors and far-field voice to its $14,000 humanoid — the platform refines itself toward real work.

會轉頭、更耐熱、只要 1.4 萬美元：宇樹 G1+ 讓人形機器人走向實用

宇樹科技發表 G1+ 人形機器人：新增 2 自由度頸部、肩部扭矩提升 110%、發熱降低 72%，加上遠場語音互動，標準版售價 9.5 萬人民幣（約 1.4 萬美元）。

First Model to Beat the Human Baseline on Every Drone Task: GPT-6 Astra's Andon Labs Sweep

On Andon Labs' Vending-Bench 2, GPT-6 Astra finished six simulated business years averaging $15,515 — nearly 3x Claude Fable 5.1 — and on Drone-Bench it became the first model to top the human-AI baseline on all five surveillance subtasks, from 3D reconstruction to following a person through an office.

首個在所有無人機任務超越人類基準的模型：GPT-6 Astra 橫掃 Andon Labs 兩大基準

在 Andon Labs 的 Vending-Bench 2 上，GPT-6 Astra 六次模擬經營年平均存入 $15,515，接近 Claude Fable 5.1 的三倍；在 Drone-Bench 上，它更成為首個在全部五項監控子任務（從 3D 環境重建到跟蹤特定人物）超越人類基準的模型。

From $12.7B to $20B in Six Months: Shield AI Is in Talks for a Round That Would Crown Defense Tech's Next Decacorn

The Information reports Shield AI is negotiating a raise at a valuation of at least $20 billion — nearly double its March price — as its Hivemind autonomy software heads into production for the Air Force's Collaborative Combat Aircraft.

半年估值翻倍再衝 200 億美元：Shield AI 正洽談新一輪融資，國防科技下一隻十角獸呼之欲出

據 The Information 報導，Shield AI 正洽談以至少 200 億美元估值進行新一輪融資，幾乎是 3 月估值的兩倍——其 Hivemind 自主飛行軟體已進入美國空軍協同作戰飛機（CCA）的量產階段。

From Lab Bench to Battlefield Tourniquet: MIT's AI-GUIDE Wins Federal Tech-Transfer Award as Novice Study Hits 93%

MIT Lincoln Laboratory and Mass General's AI-guided vascular access device takes the Federal Laboratory Consortium's 2026 Excellence in Technology Transfer Award, with a new user-validation study showing 93% first-pass needle-insertion success for clinicians with minimal ultrasound training.

從實驗室到戰場止血帶：MIT 的 AI-GUIDE 獲聯邦技術轉移大獎，新手操作成功率达 93%

MIT 林肯實驗室與麻省總醫院合作的 AI 導引血管通路裝置，榮獲聯邦實驗室聯盟 2026 年技術轉移卓越獎；最新使用者驗證研究顯示，超音波經驗有限的醫護人員操作下，模擬穿刺成功率高達 93%。

Ten Days From Ask to Workstream: China's Data Regulator Moves to Write Embodied-AI Data Standards

China's National Data Administration will develop standards for embodied-AI training data, ten days after seven robotics firms asked for them — aiming at a 10-million-hour data gap.

從請求到政策只要十天：中國國家數據局著手制定具身智慧數據標準

中國國家數據局宣布將制定具身智慧訓練數據標準，距離七家機器人企業提出相同請求僅十天——目標直指千萬小時的數據缺口。

'It Won't Be Easy': Hyundai Executive Says Boston Dynamics IPO Is Unlikely in 2027

A senior Hyundai Motor Group executive tells Reuters a Boston Dynamics IPO is unlikely next year, as the Atlas maker remains unprofitable with nearly 1.7 trillion won in cumulative losses and no at-scale deployment yet.

「沒那麼容易」：現代汽車高層透露 Boston Dynamics 明年 IPO 機率不高

路透社獨家報導，現代汽車集團高層表示 Boston Dynamics 明年不太可能上市：Atlas 尚未規模化部署、累虧近 1.7 兆韓元，估值預估從 50 兆到 100 兆韓元不等。

When 'Pretty Close' Isn't Good Enough: MIT's HardFlow Makes Generative AI Safe for the Real World

MIT researchers unveil HardFlow, a plug-and-play algorithm that enforces non-negotiable hard constraints on pretrained generative models at deployment time — no retraining required.

當「差不多」還不夠：MIT 的 HardFlow 讓生成式 AI 安全走進真實世界

MIT 研究團隊發表 HardFlow 演算法，可在部署階段直接為預訓練生成模型加上不可違反的硬約束，無需重新訓練。

Seven Million Cars as a Learning Machine: Hyundai Puts Its Data Flywheel Into Full Operation

At its Autonomous Driving Media Day, Hyundai Motor Group declared its Data Flywheel fully operational, targeting Level 2+ production with NVIDIA in H1 2028, proprietary Atria AI Level 2++ by 2029, and data volume surpassing Tesla by 2033.

七百萬輛車就是一台學習機器：現代汽車集團 Data Flywheel 正式全面啟動

在 Autonomous Driving Media Day 上，現代汽車集團宣布 Data Flywheel 全面投入運轉，目標 2028 上半年攜手 NVIDIA 量產 Level 2+，2029 下半年推出自研 Atria AI 的 Level 2++ 車款，並預估 2033 年數據累積量超越 Tesla。

From $60M to a $500M Valuation in One Summer: Sequoia Leads the Land Grab for Mecka AI's Robot Training Data

Three months after raising $60M, human-motion-data startup Mecka AI is nearing a Sequoia-led round at a ~$500M valuation — the strongest sign yet that physical-world data has become robotics' most contested resource.

一個夏天從 6,000 萬美元估值衝上 5 億：Sequoia 領投，機器人訓練資料成了 AI 下一場搶奪戰

募資 6,000 萬美元才三個月，人類動作資料新創 Mecka AI 就傳出將以約 5 億美元估值完成 Sequoia 領投的新一輪——實體世界資料已成為機器人產業最搶手的戰略資源。

From Demos to $100M in Ten Months: Skild AI Declares the Era of Robot Deployment Open

Skild AI has crossed $100M in annual revenue run rate just ten months after its first commercial deployment — 60+ customers, hundreds of robots, and a generalist brain that learns new tasks from a single video.

從 Demo 到 1 億美元只花十個月：Skild AI 宣告機器人部署時代來臨

Skild AI 在首次商業部署後十個月達成 1 億美元年營收跑速率——超過 60 家客戶、數百台機器人，靠的是一支影片就能學會新任務的通用機器人大腦。

3 Million Robots, 1 Million Autonomous Vehicles, 100,000 Drones: JD.com Goes All-In on Physical AI

At JDDiscovery 2026 in Beijing, JD.com launched a Physical AI Acceleration Plan targeting six 'world's largest' milestones, anchored by JD Logistics' plan to procure 3 million robots over five years.

300 萬機器人、100 萬自駕車、10 萬無人機：京東全面押注實體 AI

京東於北京 JDDiscovery 2026 發布「實體 AI 加速計畫」，鎖定六個「全球最大」目標，核心是京東物流五年內採購 300 萬台機器人的藍圖。

India's First Lights-Out Factory Lab: TCS Builds a Fully Robotic Battery Line in Pune to Sell the Factory of the Future

TCS has opened India's first lights-out factory lab at Sahyadri Park in Pune — a fully robotic battery pack assembly line wrapped in digital twins, vision AI and sensor-to-cloud intelligence, built to de-risk AI-first manufacturing for clients.

印度首座「關燈工廠」實驗室：TCS 在浦那打造全機器人電池產線，把未來工廠當成產品來賣

TCS 於浦那 Sahyadri Park 園區啟用印度首座關燈工廠（lights-out factory）實驗室——一條整合數位孿生、視覺 AI 與感測上雲的全機器人電池組裝產線，目標是為客戶降低 AI 優先製造的部署風險。

Beijing Pulls Up the IPO Drawbridge: CSRC 'Window Guidance' Hits Humanoid-Robot Listings After Unitree's Crash

China's securities regulator is informally telling banks and startups that humanoid-robot IPOs now require proven recurring revenue and a path to profitability — a direct response to Unitree's 5x debut pop and 45% collapse.

北京拉起 IPO 吊橋：宇樹崩跌後，證監會「窗口指導」收緊人形機器人上市

中國證監會以非正式的「窗口指導」通知投行與企業：人形機器人 IPO 如今必須證明有經常性收入、虧損持續收斂——直接回應宇樹上市首日暴漲逾 5 倍又崩跌 45% 的教訓。

From Dance Floor to War: China's PLA Is Fast-Tracking Humanoid Robots for Combat

從舞台到戰場：中國解放軍加速將人形機器人推向實戰

From Simulating the World to Acting in It: HiDream's O1-Embodied Tops RoboColiseum's Hardest Leaderboard

HiDream.ai launches HiDream-O1-Embodied, an embodied world model that ranks No. 1 on RoboColiseum's Robustness leaderboard (0.692) and pairs a 100x generative data flywheel with Noitom motion capture.

從模擬世界到親手操作：HiDream 推出 O1-Embodied，奪下 RoboColiseum 最難排行榜榜首

HiDream.ai 發布具身世界模型 HiDream-O1-Embodied，以 0.692 分登上 RoboColiseum 魯棒性排行榜第一名，並以 Noitom 動捕資料結合 100 倍生成式擴增打造資料飛輪。

Claude as Puppeteer: MIT's Phillip Isola Says Cloud LLMs Are About to Take Over Robot Bodies

In a September 7 essay, MIT's Phillip Isola argues that frontier LLMs like Claude, Fable and Astra are becoming competent 'robot-use agents' — and that any internet-connected robot could gain AI abilities overnight, with nothing more than a software update.

Claude 當提線木偶師：MIT 學者 Isola 宣稱雲端 LLM 即將接管機器人的身體

MIT 副教授 Phillip Isola 在 9 月 7 日的短文中指出，Claude、Fable、Astra 等前沿 LLM 正成為勝任的「機器人使用代理」——任何連網機器人都可能靠一次軟體更新，在一夜之間獲得 AI 能力。

Too Complex for Humans, Solvable for Machines: Arm's Rene Haas Says AI Will Cure Cancer

Arm's CEO tells the BBC that AI will find cancer cures humans cannot, humanoid robots arrive within five years, and the real bottleneck is chips — with Meta's $2bn Arm AGI demand 'off the charts'.

人類解不開、機器能解開：Arm 執行長 Haas 宣稱 AI 將治癒癌症

Arm 執行長向 BBC 表示，AI 將找出人類無法發現的癌症療法、五年內人形機器人將普及，而真正的瓶頸是晶片——Meta 的 Arm AGI 晶片需求「爆表」，已超過 20 億美元。

AI Will Cure Cancer, Says Arm's CEO — If the Chip Supply Holds

Arm CEO Rene Haas tells the BBC that AI will crack cancer's complexity in our lifetimes and put humanoid robots to work within five years — but warns chip shortages are the real bottleneck.

Arm 執行長：AI 將治癒癌症——前提是晶片供應跟得上

Arm 執行長 Rene Haas 接受 BBC 專訪時表示，AI 有生之年將破解癌症的複雜性、五年內人形機器人將廣泛應用，但晶片短缺才是真正的瓶頸。

No Teleop, No Script: Unitree's UnifoLM-X2-1.0 Drives Fully Autonomous Humanoid Combat With a Real-Time World Model

Unitree claims UnifoLM-X2-1.0 is the first real-time world-model system to control fully autonomous humanoid combat — no teleoperation, no choreography. Inside the claim, the WMA-0 lineage, and what's still unproven.

無遙控、無腳本：宇樹 UnifoLM-X2-1.0 用即時世界模型驅動全自主人形機器人格鬥

宇樹科技宣稱 UnifoLM-X2-1.0 是首個以即時世界模型驅動全自主人形機器人格鬥的系統——沒有遙控、沒有預編排動作。深入解析這項宣告、WMA-0 的技術血統，以及尚待驗證的疑點。

From Voice Commands to a Household Agent: Baidu's Xiaodu Takes Its 'Family AI Brain' to Hardware

At its September 8 launch event in Beijing, Baidu's Xiaodu unveiled an agent-first hardware lineup — Tiantian companion screens, smart displays, speakers, and cameras running the agentic 'Chaoneng Xiaodu' assistant and a 2.0 AI caregiving agent.

從語音指令到家庭智能體：百度小度把「家庭 AI 大腦」裝進硬體

百度小度於 9 月 8 日北京發表會推出以智能體為核心的硬體陣容——添添閨蜜機、智慧螢幕、音箱與攝影機，全面搭載智能體化的「超能小度」，攝影機並迎來第二代 AI 看護智能體。

From $1B to $5B in Six Months: Ukraine's UForce Courts a $500M Round to Build NATO's Drone Stack

Bloomberg reports UForce — the London-based consolidator of Ukraine's Magura sea drones and Nemesis bombers — is in talks for a ~$500M round at a ~$5B valuation led by Valor Equity Partners, a five-fold jump in six months as European defense money chases battlefield-proven autonomy.

六個月從 10 億變 50 億美元：烏克蘭 UForce 洽談 5 億美元融資，要打造北約的無人機作戰堆疊

彭博報導，整合烏克蘭 Magura 海上無人機與 Nemesis 轟炸無人機團隊的倫敦公司 UForce，正洽談由 Valor Equity Partners 領投、估值約 50 億美元的 5 億美元輪次——半年內估值翻漲五倍，歐洲國防資金正瘋狂追逐經過實戰驗證的自主系統。

Unfinished Business: Kalanick's Atoms Is Building Robotaxi Tech With Uber's $100M Quietly in the Round

An FT investigation reveals Travis Kalanick's Atoms is developing robotaxi technology under Anthony Levandowski, with Uber holding a $100M stake — the founder's return to the industry that removed him.

未竟之業：Kalanick 的 Atoms 悄悄打造 Robotaxi 技術，Uber 的 1 億美元就藏在股東名冊裡

FT 調查揭露 Travis Kalanick 的新創 Atoms 正在 Anthony Levandowski 主導下開發 Robotaxi 技術，而當年踢走他的 Uber，竟是默默投入 1 億美元的股東。

'Unfinished Business': FT Reveals Kalanick's Atoms Is Building Robotaxi Tech With Levandowski

A Financial Times investigation pulls back the curtain on Travis Kalanick's Atoms: robotaxi software in development, Anthony Levandowski on the payroll, and a $100M Uber check — the Uber founder's return to the market he was forced to abandon.

「未完的事業」：《金融時報》揭密 Kalanick 的 Atoms 正與 Levandowski 開發 Robotaxi 技術

《金融時報》調查報導揭開 Travis Kalanick 旗下 Atoms 的面紗：機器人計程車軟體正在開發中、Anthony Levandowski 已在團隊之中，Uber 更投資了 1 億美元——這位 Uber 創辦人重回他當年被迫離開的市場。

One Box for Storage, Security, and a Brain: Inside UGREEN's HomeAgent and the $9,999 Jetson Thor Hub That Wants to Run Your Whole Home

UGREEN's HomeAgent lineup merges a NAS, an NVR, and an on-device AI assistant — topping out with a $19,999 NVIDIA Jetson Thor T5000 hub delivering 2,070 TFLOPS of local FP4 compute.

一台機器搞定儲存、監控與大腦：UGREEN HomeAgent 與那台要價 9,999 美元的 Jetson Thor 家用中樞

UGREEN 的 HomeAgent 系列把 NAS、NVR 與裝置端 AI 助理三合一，最高階的 MasterAgent MA100 採用 NVIDIA Jetson Thor T5000，提供 2,070 TFLOPS 的本地 FP4 運算能力，建議售價 19,999 美元。

A Robot for the Hardest Room in the Building: Inside Hivebotics' $6M Series A

The Quiet Winner of IFA 2026: D-Robotics Wants to Be the Android of Home Robots

IFA's flashiest home robots — TCL's hey AiMe, Vbot's SuperDog, the xLean TR1 — all run on Sunrise AI chips from D-Robotics, the Horizon Robotics spinoff that just raised $270M to own the compute layer under embodied AI.

IFA 2026 的隐形赢家：D-Robotics 想当家用机器人的「安卓」

IFA 最受瞩目的三款家用机器人——TCL hey AiMe、Vbot SuperDog、xLean TR1——全都跑在 D-Robotics 的 Sunrise AI 晶片上。这家从地平线机器人拆分出来的公司刚募得 2.7 亿美元，要拿下具身智慧的运算层。

No Brush, No Cloth, No Cross-Contamination: Hivebotics Raises $6M to Mass-Produce the Abluo Restroom Robot

Singapore's Hivebotics closed a $6M Series A led by Vertex Ventures to move its Abluo full-restroom cleaning robot from 20+ pilot sites into volume production, betting that contactless steam, an articulated arm, and auditable cleaning can fix a job with 200-400% annual turnover.

One Robot for Aging Alone: Tuya's Doova Wants to Be the Whole Answer

At IFA 2026 in Berlin, Tuya Smart unveiled Doova, an AI home companion robot for seniors living alone that combines fall detection, 60-second emergency escalation, conversational companionship, and smart-home control in one soft-bodied device.

一臺機器人，照顧獨老生活：塗鴉智能 Doova 想一次給出全部答案

塗鴉智能（Tuya Smart）在柏林 IFA 2026 發表居家陪伴機器人 Doova，鎖定獨居長者：整合跌倒偵測、60 秒緊急通報、自然對話陪伴與智慧家庭中控於一身。

Three Months Out of Stealth, XDOF Is Already in Talks for a $1.2 Billion Series B

The robot-data infrastructure startup XDOF is in late-stage talks to raise a Series B at a $1.2 billion valuation led by 8VC, with annualized revenue already approaching $50 million and 20 customers including frontier AI labs.

離開潛行模式僅三個月，XDOF 已在洽談 12 億美元估值的 B 輪融資

機器人數據基礎設施新創 XDOF 據報正在洽談由 8VC 領投、估值約 12 億美元的 B 輪融資，年化營收已接近 5,000 萬美元，客戶包含多家前沿 AI 實驗室。

Banned in America, Dispensing in Berlin: Galbot's G1 Robot Works Real Pharmacy Shifts at IFA

Galbot's wheeled humanoid G1 handles 5,000 SKUs in a Beijing pharmacy at 99.5% pick success and runs 170+ autonomous stores across 40 Chinese cities — but a July 28 FCC Covered List ruling bars new units from the US, turning IFA Berlin into the company's European beachhead.

在美國被封殺，在柏林抓藥：Galbot G1 機器人在 IFA 現場示範真實藥局輪班

Galbot 的輪式人形機器人 G1 在北京藥局處理 5,000 種商品、撿貨成功率 99.5%，並在中國 40 多個城市營運超過 170 家無人店——但 7 月 28 日 FCC 將其硬體類別列入 Covered List，新款機型不得進入美國，IFA 柏林因此成為這家公司進軍歐洲的灘頭堡。

100,000 GPUs for the Robot in Every Home: Inside Figure's $3.5 Billion Nscale Compute Deal

Figure is committing $3.5 billion to Nscale for up to 100,000 NVIDIA Vera Rubin GPUs in Barstow, Texas — the largest dedicated compute buy yet for humanoid robot AI, with intent to scale past $6 billion.

十萬顆 GPU 打造家家戶戶的機器人：Figure 與 Nscale 35 億美元算力交易解析

Figure 承諾向 Nscale 投入 35 億美元，取得最多十萬顆 NVIDIA Vera Rubin GPU，部署於德州 Barstow——這是人形機器人 AI 迄今最大規模的專用算力採購，並計畫擴大至 60 億美元以上。

No Wheel, No Pedals, No Pass: NHTSA Opens Federal Audit Into Tesla's Cybercab Self-Certification

Hours after the first Cybercabs hit Austin streets, NHTSA opened an audit query covering ~1,000 vehicles into how Tesla self-certified a car with no steering wheel, pedals, or side mirrors — the same process that took Zoox two years to clear.

無方向盤、無踏板、也無豁免：NHTSA 對特斯拉 Cybercab 自我認證展開聯邦審計調查

首批 Cybercab 駛上奧斯汀街頭約一天後，NHTSA 即對約 1,000 輛車展開審計調查，矛頭指向特斯拉如何為一輛沒有方向盤、踏板與後視鏡的車自我認證合規——當年 Zoox 走完同一程序花了整整兩年。

"We Will Definitely Do a Humanoid": Sam Altman Confirms OpenAI Is Building Its Own Robot Body

In his cleatest hardware commitment yet, OpenAI's CEO confirmed on the Sources podcast that the company will build a humanoid robot of its own — putting the AI lab on a collision course with Tesla's Optimus, Figure, and its own former robotics partners.

「我們一定會做人形機器人」：Sam Altman 親口證實 OpenAI 將打造自有機器人硬體

OpenAI 執行長 Sam Altman 在 Sources Podcast 上明確承諾將打造自家的人形機器人，並同時發展資料中心專用的非人形機器人——這項宣示讓 OpenAI 正面迎戰 Tesla Optimus、Figure 與昔日的合作夥伴，也為機器人產業的「大腦授權」商業模式投下變數。

A Storm of Cybercabs: Tesla's Driverless Two-Seater Hits Austin Streets on Launch Day

Tesla's purpose-built Cybercab went live in Austin on September 3 — no steering wheel, no pedals, a 13-and-over age gate, and Musk's 'A Storm of Cybercabs' teasing a national deployment wave as roughly 45 units join the Texas registry.

Cybercab 風暴來襲：特斯拉無方向盤自動計程車正式上路奧斯汀街頭

特斯拉專為自動駕駛打造的 Cybercab 於 9 月 3 日在奧斯汀正式投入營運 — 沒有方向盤、沒有踏板、乘車年齡限制提高到 13 歲，約 45 輛已登錄德州車籍，馬斯克一句「A Storm of Cybercabs」預告更大規模的部署浪潮。

Uber and Wayve Launch London's First Robotaxi Service: Wayve's First Commercial Deployment Anywhere

Uber and British startup Wayve rolled out London's first commercial robotaxi service on September 3 — under 20 camera-and-radar Ford Mustang Mach-Es, a licensed operator up front, and Wayve's first revenue-generating deployment in its nine-year history.

Uber 攜手 Wayve 啟動倫敦首個 Robotaxi 服務：Wayve 成立九年來的首次商業部署

Uber 與英國新創 Wayve 於 9 月 3 日在倫敦上線首個商業化 Robotaxi 服務——不到 20 輛搭載攝影鏡頭與雷達的 Ford Mustang Mach-E，配有執照操作員隨車，也是 Wayve 九年歷史中第一個真正營收的部署。

The Self-Driving Car That Explains Itself: MIT and Motional's CW-Net Cracks the AV Black Box

Published in Nature today, CW-Net translates an autonomous vehicle's hidden reasoning into human concepts in real time — and helped safety drivers predict when a real robotaxi was about to make a mistake.

會解釋自己的自駕車：MIT 與 Motional 的 CW-Net 打開自駕系統黑箱

今日發表於《Nature》的 CW-Net，能即時將自駕車決策系統的內部推理轉譯為人類可理解的概念，並在真實 robotaxi 測試中幫助安全駕駛提前預測車輛的錯誤行為。

The $1,688 Humanoid: Nori Robotics' YC-Backed Bid to Collapse Robot Economics

YC S26 startup Nori Robotics is shipping a $1,688 bimanual wheeled humanoid from San Francisco — 19 DOF, open SDK, $350K in sales in six weeks, and a plan to turn every customer into a data source for generalist robot policies.

1,688 美元的人形機器人：Nori Robotics 獲 YC 投資，要徹底改寫機器人成本結構

YC S26 新創 Nori Robotics 正從舊金山出貨一款 1,688 美元的雙臂輪式人形機器人 — 19 自由度、開放 SDK、六週內締造 35 萬美元銷售，並計劃讓每一位客戶都成為通用機器人政策的資料來源。

Ex-Apple Face ID Engineers Triple Lyte's Valuation to $1.6B With $165M Series C

Physical-AI perception startup Lyte raised a $165M Series C led by Maverick Silicon at a $1.6B post-money valuation, just eight months after emerging from stealth — to put its LyteVision 'visual brain' into more robots, faster.

前 Apple Face ID 工程師團隊以 1.65 億美元 C 輪融資，將 Lyte 估值推升至 16 億美元

實體 AI 感知新創 Lyte 完成 1.65 億美元 C 輪融資，由 Maverick Silicon 領投，投後估值達 16 億美元——距離走出隱身模式僅八個月，目標是讓 LyteVision「視覺大腦」更快進駐更多機器人。

Uber Cuts 3,300 Jobs to Fund Its $10 Billion Robotaxi Future

Uber's largest restructuring since the pandemic eliminates 10% of its global workforce, flattens management by 20%, and redirects savings into an autonomous strategy with over $10 billion already committed to Avride, Lucid, Nuro and Rivian.

Uber 大砍 3,300 個職位，全力豪賭 100 億美元的自駕計程車未來

Uber 疫情以來最大規模重組：全球員工裁減 10%、管理職縮編 20%，省下的資源全數投入已承諾超過 100 億美元的 Autopilot 策略——Avride、Lucid、Nuro 與 Rivian 四路佈局。

World Labs Unveils Atlas: An Omni World Model That Generates, Reconstructs, and Simulates Any World

Fei-Fei Li's World Labs introduced Atlas on September 1 — a multimodal autoregressive diffusion transformer pretrained from scratch to natively handle text, images, video, and 3D, generating minute-long 1440p videos and beating specialist models at 3D reconstruction.

World Labs 發表 Atlas：能夠生成、重建與模擬任何世界的全模態世界模型

李飛飛創辦的 World Labs 於 9 月 1 日推出 Atlas——一個從零預訓練的多模態自回歸擴散 Transformer，原生支援文字、影像、影片與 3D，可生成長達一分鐘的 1440p 影片，並在 3D 重建任務上擊敗專用模型。

NavMCP Scaffolds VLMs and Navigation Models Into Physical-World Agents That Get Better the Longer the Task

A new paper from SJTU, Alibaba's Qwen team, and Peking University couples a VLM reasoning agent with a navigation foundation model executor through three protocol channels — reaching 78.3% success on a Unitree Go2 with margins that grow from 10 to 45 points as task horizons lengthen.

NavMCP：把視覺語言模型與導航基礎模型鷹架成實體世界代理，任務越長優勢越大

上海交大、阿里 Qwen 團隊與北京大學的新論文，透過意圖、觀測、記憶三個協定通道將 VLM 推理代理與導航基礎模型執行器耦合——在 Unitree Go2 上達成 78.3% 成功率，且領先幅度隨任務視野拉長從 10 分一路擴大到 45 分。

ByteDance's Lucida Turns Messy Room Videos Into Editable 3D Scenes for Robots

ByteDance Seed's Lucida pipeline parses cluttered indoor video into per-instance scene graphs, generates an asset per object, and lets a VLM policy drive the 3D editor's own gizmo handles until it decides placement is done — posting a 69% mAP gain on R2S-Scene.

ByteDance Lucida：把雜亂房間影片變成機器人可用的可編輯 3D 場景

ByteDance Seed 的 Lucida 管線把雜亂室內影片解析成逐物件場景圖、為每個物件生成資產，再由 VLM 策略 GizmoAct 直接操作 3D 編輯器的 gizmo 手柄決定擺放完成與否——在 R2S-Scene 上繳出 69% 的 mAP 增益。

Mitsubishi Electric's TUSS Gives Physical AI a Pair of Ears

A single prompt-driven AI model now handles speech separation, enhancement, and environmental sound extraction at once — aimed at factory floors and public spaces, with a live demo at CEATEC 2026.

三菱電機 TUSS：讓實體 AI 長出一對耳朵

單一提示驅動的 AI 模型，同時搞定語音分離、語音強化與環境音抽取——瞄準工廠現場與公共空間，並將在 CEATEC 2026 現場實機展演。

HUMAIN and Applied Intuition Will Build the World's Largest Autonomous Trucking Network in Saudi Arabia

Saudi Arabia's HUMAIN and Silicon Valley's Applied Intuition will deploy thousands of Level 4 autonomous trucks across the Kingdom's freight corridors by 2030, the first step of a national physical AI strategy spanning robotaxis, ports, mining, and construction.

HUMAIN 攜手 Applied Intuition，將在沙烏地阿拉伯打造全球最大自動駕駛卡車網路

沙烏地阿拉伯 HUMAIN 與矽谷 Applied Intuition 宣布策略合作，2030 年前將在王國主要物流走廊部署數千輛 Level 4 自動駕駛卡車，並以此為起點，將實體 AI 擴展到機器人計程車、港口、礦業與營造等產業。

Anthropic's Model Hardware Standard: AI Agents Take Control of the Lab Bench

Anthropic's Model Hardware Standard gives AI agents a universal interface to operate microscopes, liquid handlers, and robotic arms — turning weeks of integration work into minutes.